PJFP.com

Pursuit of Joy, Fulfillment, and Purpose

Tag: Grok

  • SpaceX S-1 IPO Filing Breakdown, Ticker SPCX on Nasdaq and Nasdaq Texas, xAI Integration, Musk’s Trillion Share Mars Pay Plan, $18.7B Revenue, and the 100 Gigawatt Orbital AI Compute Bet

    Space Exploration Technologies Corp. filed its S-1 registration statement with the SEC on May 20, 2026, kicking off the largest and weirdest IPO in modern capital markets history. The 280-page preliminary prospectus proposes to list Class A common stock on both the Nasdaq Stock Market and the new Nasdaq Texas exchange under the ticker SPCX, bundles xAI into SpaceX as a third reportable segment via a February 2026 reorganization under common control, and asks public investors to underwrite a $28.5 trillion total addressable market that explicitly includes asteroid mining, lunar manufacturing, Mars passenger transport, and 100 gigawatts per year of orbital AI compute on solar-powered satellites. The filing reports $18.67 billion of 2025 revenue and a $4.94 billion net loss, with a Q1 2026 net loss of $4.28 billion driven almost entirely by the AI segment’s $7.7 billion of quarterly capex.

    TLDR

    SpaceX is going public on Nasdaq and Nasdaq Texas as SPCX, led by Goldman Sachs, Morgan Stanley, BofA Securities, Citigroup, and J.P. Morgan. The company has been reincorporated in Texas, headquartered at Starbase, structured as a perpetual dual-class controlled company with Class B shares carrying 10 votes each and electing a majority of the board, and post-merger contains three segments: Space (Falcon, Dragon, Starship), Connectivity (Starlink with 10.3 million subscribers across 164 countries and roughly 9,600 satellites in orbit), and AI (the former xAI, including the Colossus and Colossus II superclusters in Memphis totaling about 1.0 gigawatt of nameplate compute, Grok, and the X platform with 550 million MAUs). Revenue grew from $10.4 billion in 2023 to $14.0 billion in 2024 to $18.7 billion in 2025, with Connectivity contributing $11.4 billion at a 63% segment Adjusted EBITDA margin and the new AI segment burning $1.2 billion of segment Adjusted EBITDA in 2025 while spending $12.7 billion of capex. Elon Musk holds an unspecified majority of the voting power, has a base salary of $54,080 unchanged since 2019, no key-person life insurance, and was granted in January and March 2026 a combined roughly 1.3 billion performance-restricted Class B shares that vest against market-cap milestones from $500 billion up to $7.5 trillion, with the highest tranches contingent on building a permanent Mars colony of one million inhabitants and on deploying non-Earth data centers delivering 100 terawatts of compute per year. The prospectus discloses Anthropic’s $1.25 billion per month compute deal through May 2029, a $60 billion option to acquire Cursor (Anysphere) with a $10 billion combined break fee, the Terafab one-terawatt-per-year chip JV with Tesla and Intel, the $19.6 billion EchoStar spectrum acquisition, a $20 billion SpaceX Bridge Loan, a $5 billion amended revolver, a Houston-exclusive Texas Business Court forum clause with ICC arbitration fallback, and several uniquely SpaceX risk factors including third-party Musk conduct triggering foreign asset seizures, anti-satellite weapons, cascading cyber-induced orbital debris events, and Grok’s named “Spicy” Imagine Mode and “Unhinged” Voice Mode.

    Key Takeaways

    • Ticker SPCX, dual listed on Nasdaq and Nasdaq Texas, Class A par $0.001, joint lead bookrunners Goldman Sachs, Morgan Stanley, BofA Securities, Citigroup, and J.P. Morgan, with a 22-firm syndicate including Barclays, Deutsche Bank, RBC, UBS, Wells Fargo, Allen & Company, Cantor, Needham, Raymond James, Societe Generale, Stifel, William Blair, BTG Pactual, ING, Macquarie, Mirae Asset, Mizuho, and Santander.
    • Headquartered at 1 Rocket Road, Starbase, Texas. Reincorporated from Delaware to Texas on February 14, 2024. Five-for-one forward stock split executed May 4, 2026. All share data in the filing is post-split.
    • Perpetual dual-class structure with no sunset. Class A carries 1 vote per share, Class B carries 10 votes per share, Class C carries no votes (and has been eliminated via the Class C Reclassification). Class B converts to Class A only on a non-permitted transfer.
    • Class B holders elect a majority of the board (the Class B Directors), as long as any Class B shares remain outstanding. Removing Musk from CEO or Chairman requires a separate Class B majority vote. SpaceX will be a Nasdaq controlled company and will rely on the exemptions, meaning no requirement for fully independent compensation or nominating committees.
    • Consolidated revenue: $10.39 billion in 2023, $14.02 billion in 2024, $18.67 billion in 2025, and $4.69 billion in Q1 2026 (up 15.4% year over year). Financials are retrospectively recast to combine xAI and X Holdings since both transactions were between entities under Musk’s common control.
    • Net income (loss): $(4.63) billion in 2023, $0.79 billion in 2024, $(4.94) billion in 2025, and $(4.28) billion in Q1 2026. Accumulated deficit pro forma $41.31 billion as of March 31, 2026.
    • Connectivity (Starlink) is the cash engine. 2025 revenue $11.39 billion, up 49.8%. 2025 operating income $4.42 billion, up 120.4%. 2025 segment Adjusted EBITDA $7.17 billion, up 86.2%. Consumer subscriptions are more than 60% of Connectivity revenue.
    • Starlink subscribers: 2.3 million at year-end 2023, 4.4 million at year-end 2024, 8.9 million at year-end 2025, and 10.3 million as of March 31, 2026. Roughly 9,600 broadband and mobile satellites in low Earth orbit, about 75% of all active maneuverable satellites globally. Available in 164 countries and territories.
    • Starlink ARPU is declining as the mix shifts international and lower priced: $99 monthly in 2023, $91 in 2024, $81 in 2025, $66 in Q1 2026. Management says this is expected to continue.
    • Starlink direct to cell now has roughly 650 V1 Mobile satellites and 7.4 million monthly unique devices across about 30 countries, with partnerships across roughly 30 mobile network operators including T-Mobile, Rogers, KDDI, Optus, Telstra, One NZ, Kyivstar, VMO2, Salt, and Entel. V3 satellites begin deploying in the second half of 2026, designed for 1 Tbps downlink per satellite with up to 60 per Starship launch (a 20x payload-capacity step over Falcon 9).
    • Space segment now generates lower revenue growth because Starlink dedicated launches are not booked as inter-segment revenue. Space revenue: $3.56 billion (2023), $3.80 billion (2024), $4.09 billion (2025). Falcon launches in 2025: 165 total, 43 third-party customer and 122 internal Starlink. Mass to orbit: 1,210 metric tons (2023), 1,699 (2024), 2,213 (2025). SpaceX has now launched more than 80% of the world’s mass to orbit since 2023.
    • Falcon 9 has flown roughly 620 missions with greater than 99% mission success. A single booster has been reflown 34 times. Falcon Heavy is 11-for-11 since 2018 and certified for NSSL. SpaceX flew 11 of 12 NSSL medium and heavy lift missions in 2025.
    • Starship has completed 11 flight tests and is preparing the 12th, debuting next-generation Starship, Super Heavy, and Raptor 3 from a new Starbase pad. V3 is designed for 100 metric tons fully reusable to LEO, V4 targets 200 tons. Cumulative Starship R&D investment is greater than $15 billion, including $3.00 billion in 2025 alone. Operational payload delivery to orbit is expected in the second half of 2026.
    • Dragon has flown 78 crewmembers from 20 countries since 2020 and Cargo Dragon remains the only spacecraft capable of returning meaningful mass from the ISS.
    • AI segment, the absorbed xAI business plus X, generated $818 million Q1 2026 revenue but operating losses of $(2.47) billion and segment Adjusted EBITDA of $(609) million. AI capex was $7.72 billion in Q1 2026 alone, dwarfing Space ($1.05 billion) and Connectivity ($1.33 billion).
    • Colossus and Colossus II in Memphis and Southaven Mississippi together provide about 1.0 gigawatt of nameplate compute draw. Colossus came online in 122 days with about 100,000 H100s. Colossus II added 110,000 GB200s in 91 days and 110,000 GB300s in 64 days. Next phase: another 220,000 GB300s and 400 megawatts. Industry benchmark for a 100 megawatt greenfield datacenter is two years.
    • Grok and X together have 1.3 billion supported accounts on a trailing basis, about 550 million MAUs, roughly 117 million MAUs using Grok AI features, and roughly 350 million daily posts. Imagine generates about 10 billion images and 2 billion videos per month. Paid subscribers totaled 6.3 million as of March 31, 2026 (4.4 million X Premium variants plus 1.9 million SuperGrok variants).
    • Disclosed Anthropic cloud services agreements signed May 2026: Anthropic pays $1.25 billion per month for compute capacity on Colossus and Colossus II through May 2029, ramping in May and June 2026, with 90-day termination by either party.
    • Cursor (Anysphere) compute agreement and acquisition option signed April 2026: SpaceX has the right but not the obligation to acquire Cursor at an implied $60.0 billion equity value, paid in Class A stock priced off the SPCX VWAP. SpaceX-side termination or breach triggers a $1.5 billion termination fee plus an $8.5 billion deferred services fee.
    • Terafab JV with Tesla, announced March 2026, joined by Intel in April 2026, targets one terawatt per year of compute hardware production. The filing explicitly notes that neither Tesla nor Intel is obligated to remain, and definitive agreements may not be signed.
    • Macrohard, in development with Tesla, is described as a platform designed to fully emulate digital workflows, augment human computer operation, and create a fully AI-operated software company.
    • EchoStar Spectrum Transaction (AWS-3, AWS-4, H-block, 65 megahertz US plus global MSS) was FCC-approved May 12, 2026. Total deal value $19.6 billion, including roughly $11.1 billion of equity (261.8 million Class A shares at an implied $42.40) and up to $8.5 billion of debt assumption. Closing expected around November 30, 2027.
    • Balance sheet as of March 31, 2026: cash and equivalents $15.85 billion, short-term marketable securities $7.82 billion, total assets $102.09 billion, total liabilities $60.51 billion, total debt principal $29.13 billion. The $20 billion SpaceX Bridge Loan (Goldman Sachs Bank USA as administrative agent, March 2026) refinanced legacy X and xAI debt and must be repaid within six months of IPO. The amended SpaceX Credit Facility, also May 2026, was upsized to $5.0 billion and extended to May 19, 2031.
    • Use of proceeds: expansion of AI compute infrastructure, enhancements to launch infrastructure and launch vehicles, increases in satellite constellation scale and capacity, and general corporate purposes. No dividends are anticipated and the credit agreements restrict them.
    • Total addressable market estimate of $28.5 trillion (ex-China and Russia): Space $370 billion, Connectivity $1.6 trillion ($870 billion broadband and $740 billion mobile), and AI $26.5 trillion ($2.4 trillion infrastructure, $760 billion consumer subscriptions, $600 billion digital advertising, and $22.7 trillion enterprise applications).
    • Stated future markets explicitly listed in the prospectus: point-to-point Earth transport via Starship, space tourism, in-orbit manufacturing including pharmaceuticals and materials, passenger and cargo to Moon and Mars, lunar mining of rare materials, lunar mass driver, lunar factories building AI compute satellites, asteroid mining, and orbital solar-powered AI. The headline aspirational target is 100 gigawatts per year of orbital AI compute on solar-powered satellites in Sun-synchronous orbit, with first deployments targeted as early as 2028.
    • Musk 2025 total compensation $54,080 (base salary unchanged since 2019, tied historically to California’s exempt-employee minimum). No bonus, no stock or option awards reported for 2025. SpaceX maintains no key-person life insurance on Musk.
    • January 13, 2026 Musk grant: 1 billion performance-based restricted Class B shares across 15 equal tranches tied to market-cap milestones from $500 billion to $7.5 trillion (in $500 billion increments), with at least one tranche additionally gated on “a permanent human colony on Mars with at least one million inhabitants” and on continued employment.
    • March 23, 2026 Musk replacement award (assumed from xAI): 302,072,285 performance-based restricted Class B shares across 12 tranches from $1.065 trillion to $6.565 trillion market cap, additionally requiring completion of “non-Earth-based data centers capable of delivering 100 terawatts of compute per year.” Replaces an earlier xAI award after Musk had already earned and canceled 25,172,695 Class A shares at the first milestone.
    • Gwynne Shotwell 2025 total compensation $85.81 million, primarily option awards. Bret Johnsen (CFO) 2025 total compensation $9.84 million. Non-employee directors received zero cash and zero equity for 2025 service.
    • Board of 8 post-IPO: Musk (Chairman, CEO, CTO), Shotwell (President, COO), Antonio Gracias (Valor Management), Ira Ehrenpreis (DBL Partners and Tesla), Randy Glein (DFJ Growth, audit chair), Donald Harrison (Google), Steve Jurvetson (Future Ventures), and Luke Nosek (Gigafund and Founders Fund). Class B Directors: Musk, Shotwell, Gracias, Harrison, Nosek. Common Stock Directors: Ehrenpreis, Glein, Jurvetson.
    • Lock-up is 180 days for company, directors, and officers, but Musk and certain significant investors are subject to an extended 366-day lock-up, and 100% of Musk’s shares are explicitly not subject to early-release tiers. A Directed Share Program with Schwab, Fidelity, Robinhood, SoFi, and E*TRADE handles retail allocation; DSP shares have no lock-up.
    • Corporate Opportunities waiver in the charter renounces interest in business opportunities presented to directors, officers, board observers, and their affiliates. Musk and his affiliates are explicitly not restricted from competing with SpaceX. This carve-out covers Tesla, Neuralink, The Boring Company, and any future Musk venture.
    • Exclusive forum is the Texas Business Court, Eleventh Division, in Houston, including for federal securities claims. If unenforceable, the fallback is mandatory ICC arbitration in Houston under Expedited Procedure Rules. Jury trial is waived. Class actions are prohibited.
    • Texas Business Organizations Code carve-outs: Section 21.419 codifies a statutory business-judgment-rule presumption, Section 21.552 requires 3% minimum ownership to bring derivative proceedings, and Section 21.373 (2025) requires 3% ownership for six months plus solicitation of 67% of voting power for shareholder proposals (SpaceX concedes enforceability is “expected” to be challenged).
    • Unprecedented risk-factor disclosure: in August 2024 Brazil’s Supreme Court froze Starlink’s Brazilian assets over the conduct of X “when X was not owned by us and was only affiliated with Mr. Musk.” SpaceX warns that third-party Musk conduct may continue to trigger foreign retaliation against SpaceX.
    • Risk language names Grok’s “Spicy” Imagine Mode and “Unhinged” Voice Mode as carrying heightened risks of explicit content, misinformation, and “potential nonconsensual or exploitative imagery.” A putative class action over content “representing children in sexualized contexts” is disclosed, as is an Irish DPC GDPR inquiry into Grok and an FTC inquiry into chatbots as companions for children and teens.
    • The S-1 uses the term “Department of War” (not Defense) for the federal customer requiring CMMC compliance and discloses that anti-satellite weapons have been publicly discussed by foreign governments as a tool against the Starlink constellation. A cyberattack-induced cascading Kessler-style debris event is cited as a possibility.
    • Workforce of more than 22,000 full-time employees globally, with no collective bargaining and engineering acceptance rate under 2% in 2025.
    • Operating asset footprint: Starbase (Texas, HQ, Starship), Hawthorne (California, Falcon, Dragon, Merlin and Raptor), McGregor (Texas, engine testing), Redmond (Washington, Starlink satellite production at about 70 per week), Bastrop (Texas, terminal production at tens of thousands per day, doubling in 2026 to include AI compute satellites), Kennedy and Cape Canaveral (Florida, LC-39A, SLC-40, SLC-37 in build for Starship), Vandenberg (California, SLC-4 polar launches), Memphis and Southaven (Tennessee and Mississippi, Colossus data centers), Palo Alto (California, xAI HQ), more than 400 Starlink ground stations globally, and three autonomous spaceport drone ships including “Of Course I Still Love You,” “Just Read the Instructions,” and “A Shortfall of Gravitas.”
    • Related party transactions of note: roughly $20.2 billion of equipment lease undiscounted payments to Valor (Gracias) entities guaranteed by SpaceX; aircraft, security, and tunnel-construction payments to Musk affiliates; xAI subsidiary leases real property from Musk Industries LLC.
    • Pampena v. Musk: an April 3, 2026 partial judgment in the Northern District of California, where a jury found Musk personally violated Section 10(b) and Rule 10b-5 on two May 2022 statements regarding his Twitter purchase. Post-trial motions are pending. The 2018 SEC “funding secured” settlement is also disclosed.
    • Critical accounting policy quirks: flight vehicles are depreciated over expected average number of flights rather than time. Starship costs are expensed to R&D until commercialization, then capitalized. Starlink dedicated launch costs are capitalized into Connectivity PP&E rather than booked as inter-segment Space revenue, which mechanically suppresses the headline Space growth rate.
    • The One Big Beautiful Bill Act (Public Law 119-21) reversed a $659 million U.S. R&D credit deferred tax asset recognized in 2024, driving the 2025 income tax provision of $718 million versus a $549 million benefit in 2024.
    • Pre-IPO ownership pro forma at March 31, 2026: Class A 6,824,581,339 shares and Class B 5,695,729,430 shares outstanding, for a combined 12.52 billion shares before primary issuance. Class C and the redeemable convertible preferred are converted/reclassified at close.
    • Authorized capitalization post-IPO: 36.13 billion Class A, 6.13 billion Class B, 10.0 billion Class C (none issued), and 2.4 billion preferred (none issued). Headroom for future issuance is enormous.
    • Five-for-one stock split executed May 4, 2026 to set the IPO share count and round-lot price. Price range, share count, and proceeds are bracketed in this preliminary filing and will be updated before launch.

    Detailed Summary

    A different kind of S-1 from the start

    Most S-1 filings open with corporate prose and a careful, neutral business description. SpaceX opens with an Elon Musk epigraph about wanting to wake up in the morning and “think the future is going to be great,” a mission statement that says the company exists “to make life multiplanetary, to understand the true nature of the universe, and to extend the light of consciousness to the stars,” and a Kardashev Type II framing that treats the next century of capital allocation as a civilizational project. Investors are being told, in legally binding language, that single-planet existence is “a single point of failure” and that the company is hedging against humans sharing the fate of the dinosaurs. The filing dual-lists SPCX on Nasdaq in New York and Nasdaq Texas in Dallas, picks the new Texas Business Court in Houston as exclusive forum, and reincorporates from Delaware to Texas. Every macro signal is set deliberately.

    Three segments after the xAI absorption

    The most consequential mechanical change in the S-1 is the retrospective recast of financial statements to combine xAI Holdings and X Holdings into SpaceX. Both transactions are accounted for as reorganizations of entities under common control (Musk’s), so prior-period revenue, opex, and capex move into the SpaceX line items rather than appearing as acquired-business additions. This is what produces the headline numbers: $10.4 billion (2023), $14.0 billion (2024), $18.7 billion (2025). The Space segment includes Falcon, Dragon, and Starship. Connectivity is Starlink in all its consumer, enterprise, government, and mobile forms plus the Starshield military variant. AI is the former xAI in full: Colossus and Colossus II superclusters, Grok, the X platform, and the Imagine media products. The recast also explains why net income flips so violently year to year. 2024’s $791 million net income reflects a quieter pre-merger SpaceX. 2025’s $4.94 billion net loss and Q1 2026’s $4.28 billion loss reflect the integrated AI business burning capital at unprecedented rate.

    Connectivity is the cash engine

    Starlink is the only segment that looks like a normal high-margin growth business. Revenue rose 96.4% in 2024 and another 49.8% in 2025 to $11.39 billion. Operating income tripled in 2024 and then doubled again in 2025 to $4.42 billion. Segment Adjusted EBITDA in 2025 was $7.17 billion, an EBITDA margin north of 60%. Subscribers grew from 2.3 million to 10.3 million in twenty-seven months. The constellation is now roughly 9,600 satellites, about 75% of all active maneuverable satellites on orbit. Inter-satellite laser links exceed 23,000, forming a mesh that delivers 700+ Tbps of cumulative downlink. ARPU is declining steadily, from $99 monthly in 2023 to $66 in Q1 2026, but management frames this as deliberate international mix shift toward lower priced plans and notes that direct-to-cell is just beginning to monetize. Roughly 650 V1 Mobile satellites already provide service to 7.4 million monthly unique devices through partnerships with roughly 30 mobile network operators. The EchoStar spectrum acquisition adds 65 megahertz in the US plus global MSS spectrum to support V2 Mobile broadband and 5G IoT starting in 2027.

    Space economics are obscured by accounting

    The Space segment looks small in the headline financials ($4.09 billion of 2025 revenue, an operating loss of $657 million) until you understand the accounting. Starlink launches are capitalized into Connectivity PP&E rather than booked as inter-segment Space revenue. That single policy is why 2025 Space revenue grew only 7.6% even though SpaceX flew 170 missions, of which 122 were internal Starlink. The actual operating reality is that SpaceX flew more than 80% of the world’s mass to orbit in 2025, owns 24 flight-proven reusable Falcon 9 boosters certified for 40 flights each, has refln a single booster 34 times, and has invested more than $15 billion in Starship to date. Starship’s eleventh flight test is on the books, the twelfth will debut the next-generation vehicle and Raptor 3 engine, and operational payload delivery to orbit is targeted for the second half of 2026. V3 Starship is designed to deliver 100 tons to LEO fully reusable and to carry up to 60 V3 Starlink satellites per launch, a 20x payload step over Falcon 9. The Starship cost target is a 99% reduction against the historical $18,500 per kilogram average, on the way to “airline-like” reflight cadence.

    AI is a money furnace with a thesis

    The AI segment is brand new to the SpaceX line item set and dominates the loss line. AI generated $3.20 billion of 2025 revenue (up 22.2%) but lost $6.36 billion at the operating line, much of it driven by GPU depreciation. AI capex was $12.73 billion in 2025 and another $7.72 billion in Q1 2026 alone. Colossus came online in 122 days with about 100,000 H100s and 130 megawatts. Colossus II followed with 110,000 GB200s in 91 days and 110,000 GB300s in 64 days, with another 220,000 GB300s and 400 megawatts in the next phase. The two superclusters now draw about one gigawatt combined. Grok-5 is training on Colossus II, targeting multi-trillion parameters. The X platform contributes 550 million MAUs and roughly 350 million daily posts to the segment, with 117 million MAUs touching Grok AI features. The thesis the prospectus is pitching is vertical integration on physics: SpaceX controls power generation (data center turbines and, eventually, orbital solar), launch (Starship to lift orbital compute satellites), satellite manufacturing (Redmond and Bastrop), chip supply (Terafab JV with Tesla and Intel for one terawatt per year of compute hardware), and the application layer (Grok and X). Management calls this “shovels-to-tokens” and argues no other AI company has this much control over the physical stack.

    The Anthropic, Cursor, and Terafab carve-outs

    Three subsequent events disclosed in the S-1 reframe SpaceX as a cloud and software platform as much as a hardware company. Anthropic signed cloud services agreements in May 2026 to pay $1.25 billion per month for Colossus and Colossus II capacity through May 2029, ramping in May and June 2026. The Cursor (Anysphere) agreement signed April 2026 includes both a compute commitment and an option for SpaceX to acquire the company at a $60 billion implied equity value, with a $1.5 billion termination fee and an $8.5 billion deferred services fee if SpaceX breaches or terminates. Terafab is a manufacturing JV with Tesla, joined by Intel in April 2026, with a stated one terawatt per year compute hardware production target. The prospectus is explicit that Tesla and Intel are not obligated to remain in Terafab and that no definitive agreements may be signed. Anthropic, the leading commercial competitor to OpenAI, is now SpaceX’s largest disclosed cloud customer.

    The Musk pay package

    The CEO compensation disclosure is the most aggressive in S-1 history. Musk’s reported 2025 total compensation was $54,080, a base salary unchanged since 2019. SpaceX maintains no key-person life insurance on him. Then on January 13, 2026 the board granted him one billion performance-based restricted Class B shares, vesting across fifteen equal tranches as market capitalization milestones are achieved at $500 billion increments from $500 billion all the way to $7.5 trillion, with at least one tranche additionally conditioned on the existence of a permanent human Mars colony of at least one million inhabitants and on continued employment. On March 23, 2026 the board granted an additional 302.07 million performance-based restricted Class B shares across twelve tranches from $1.065 trillion to $6.565 trillion of market cap, additionally requiring the completion of “non-Earth-based data centers capable of delivering 100 terawatts of compute per year.” This second grant replaces an earlier xAI award after Musk had already earned 25.17 million Class A shares at the first xAI milestone, which were then canceled and rolled in. The combined package is roughly 1.3 billion restricted Class B shares, dwarfing the Tesla 2018 award that previously held the record. Other executive comp is more conventional. Gwynne Shotwell’s 2025 total was $85.81 million, primarily option awards. Bret Johnsen, CFO, received $9.84 million. Non-employee directors received zero cash and zero equity for 2025 service.

    Governance built to be Musk-proof in one direction only

    SpaceX takes the dual-class playbook further than any prior tech IPO. Class B carries 10 votes per share, has no sunset, and elects a majority of the board as a separate class. Removing Musk from CEO or Chairman requires a separate Class B majority vote, and Musk holds the majority of Class B. The charter renounces interest in business opportunities presented to Musk and his affiliates, explicitly preserving his right to run competing ventures (Tesla, Neuralink, The Boring Company, anything next). The company opts into the Texas Business Organizations Code’s Section 21.419 business-judgment-rule presumption, requires 3% ownership to bring a derivative suit, requires 3% ownership for six months plus solicitation of 67% of voting power to bring shareholder proposals under Section 21.373 (a provision SpaceX itself concedes will likely be challenged in court), picks the Texas Business Court in Houston as exclusive forum even for federal securities claims, and falls back to mandatory ICC arbitration in Houston with Expedited Procedure Rules if forum exclusivity is struck down. Jury trials are waived. Class actions are prohibited. SpaceX will be a controlled company and will rely on Nasdaq exemptions from independent committee requirements. Musk and certain significant investors are subject to a 366-day lock-up rather than the standard 180 days, and 100% of Musk’s shares are excluded from the early-release tiers other holders enjoy.

    Risk factors disclose things no S-1 has disclosed before

    The Risk Factors section contains language no prior S-1 has used. SpaceX warns that “actions and statements of Mr. Musk and his affiliated ventures, whether or not directly relating to us, may draw significant public attention and scrutiny” and notes that in August 2024 the Brazilian Supreme Court froze Starlink’s Brazilian assets over the conduct of X “when X was not owned by us and was only affiliated with Mr. Musk.” That is the precedent: a foreign government seized SpaceX assets over Musk’s separate business conduct. The filing names Grok’s “Spicy” Imagine Mode and “Unhinged” Voice Mode as carrying heightened risks of explicit content and “potential nonconsensual or exploitative imagery,” discloses a putative class action over content “representing children in sexualized contexts,” an Irish DPC GDPR inquiry into Grok’s processing of EU children’s data, and an FTC inquiry into chatbots as companions for children and teens. The orbital risk language describes a cyberattack-triggered cascading Kessler-style debris event that could render SpaceX-licensed orbits “unusable for an extended period,” notes that “certain foreign governments have publicly discussed the potential use of anti-satellite weapons against the Starlink constellation,” and acknowledges that the FAA does not currently permit return-to-launch-site reentries for Starship and the company will require a waiver “which is not guaranteed.” The filing also uses “Department of War” rather than “Department of Defense” when discussing CMMC compliance for federal customers, reflecting the recent rebranding.

    Capital position and the bridge loan time bomb

    The balance sheet is large but the debt structure tells a story about why an IPO is urgent now. SpaceX has $15.85 billion of cash and $7.82 billion of short-term marketable securities against total debt principal of $29.13 billion. The largest piece is the $20 billion SpaceX Bridge Loan signed March 2026 with Goldman Sachs Bank USA as administrative agent, used to refinance legacy X and xAI debt (including X B-1, X B-3, and xAI 12.5% Senior Secured Notes). The bridge matures September 2, 2027 (extendable to March 2028 with a 0.25% fee per quarter), priced at Term SOFR plus 0.75% to 1.75%, with 0.125% duration fees kicking in at year one. It must be repaid within six months after IPO completion. The amended SpaceX Credit Facility was upsized to $5.0 billion and extended to May 19, 2031 in May 2026, with a $2.0 billion performance LC sublimit. The leverage covenant is 3.75x maximum (4.25x post-qualified acquisition). Capex is enormous and consistent: $20.74 billion in 2025 ($3.83 billion Space, $4.18 billion Connectivity, $12.73 billion AI), $10.11 billion in Q1 2026 alone. Operating cash flow ($6.79 billion in 2025) does not cover capex, and the gap is being filled by financing activity ($26.35 billion of net financing inflow in 2025).

    The 100 gigawatt orbital AI bet

    Buried in the Business section is the future-markets framing that justifies the AI-segment burn rate. SpaceX is asking public investors to underwrite a plan to deploy 100 gigawatts per year of orbital AI compute on solar-powered satellites in Sun-synchronous orbit. Reaching that scale requires thousands of Starship launches per year and roughly one million metric tons of mass to orbit annually. First modular orbital AI shells are targeted for “as early as 2028.” The justification given is that the Sun contains roughly 99.8% of the solar system’s energy, that orbital compute escapes terrestrial constraints on power, cooling, latency, and permitting, and that no other AI company controls the physical stack required to deploy at that scale. The prospectus stitches this directly to the Mars project: lunar mining of rare materials, lunar mass drivers to launch satellites at low cost, and lunar factories building AI compute satellites are listed alongside asteroid mining and Mars passenger transport as the future markets investors are being asked to value. The risk language acknowledges that none of these markets currently exist and that breakthrough advances in nuclear energy could moot the orbital compute thesis entirely. Investors are being asked to take Musk’s word that the long-tail outcomes are real options.

    Thoughts

    The most important number in this S-1 is not the revenue, the loss, or the implied valuation. It is the $54,080 Musk salary unchanged since 2019 against the 1.3 billion performance-restricted Class B shares contingent on a Mars colony and 100 terawatts of off-Earth compute. This is a pay package that resolves the question of whether SpaceX is a public-markets-style optimized corporation by answering it directly: no. SpaceX is going public on Musk’s terms, with a perpetual dual-class structure, a controlled-company exemption, a Houston exclusive forum, an arbitration backstop, a class-action prohibition, a charter that explicitly renounces interest in business opportunities Musk gets pitched elsewhere, and a CEO compensation structure that pays nothing for normal performance and 1.3 billion shares for an interplanetary civilization. Investors who buy SPCX are not buying voting power. They are buying optionality on the most ambitious capital allocation thesis a public company has ever attempted, contingent on Musk continuing to deliver outcomes the rest of the industry cannot.

    The xAI absorption is the most consequential corporate event in the prospectus and the one most worth scrutinizing. Accounting it as a common-control reorganization is technically defensible because Musk controlled all three entities, but the practical effect is to fold xAI’s enormous compute burn and X’s separate litigation surface area into SpaceX’s reported financial history without showing the deals as acquisitions. The Q1 2026 net loss of $4.28 billion is almost entirely xAI capex pulling forward. The two segments that actually make money (Connectivity at a 63% Adjusted EBITDA margin, Space when you adjust for the launch accounting policy) are being asked to subsidize an AI build-out that requires the orbital compute thesis to come true to ever generate adequate returns. Strip out AI and SpaceX would be one of the highest-quality businesses ever taken public. Include AI and it is something more like a venture-stage company stapled to a cash-flow machine, with the venture stage absorbing the cash. That is the trade the IPO is asking the market to price.

    The risk-factor language about third-party Musk conduct triggering foreign asset seizures is the cleanest single articulation in any S-1 of why founder-led companies with cross-portfolio exposure are different from normal public companies. The Brazil precedent is real, the legal theory is established, and the prospectus admits it directly. Buying SPCX means accepting that a fight between Musk and a foreign government over X content moderation, a Neuralink ethics dispute, a Boring Company permit fight, or a future venture entirely unrelated to space could trigger a freeze on Starlink subscriber revenue in that country. The Corporate Opportunities waiver is the legal mechanism that makes this acceptable to the board. It is far from clear that it is acceptable to public-market shareholders. The early reception of SPCX will partly be a referendum on whether the market thinks Brazil 2024 was a one-time event or a template.

    The Anthropic disclosure is the funniest detail. SpaceX, controlled by Musk, is now selling roughly $15 billion per year of compute to Anthropic, a company explicitly founded by former OpenAI researchers who broke away from the OpenAI-Musk faction in 2021. SpaceX-Colossus is now Anthropic’s largest disclosed compute supplier through May 2029, on 90-day termination by either side. The OpenAI lawsuit, the xAI launch, and the Grok positioning as the “truth-seeking” anti-OpenAI all sit in tension with the fact that Anthropic now anchors xAI’s third-party compute revenue. The economic logic is simple. The political logic, given the lockup of compute supply that this deal effectively creates, is fascinating. Public investors are being asked to underwrite a business where the largest compute customer is a direct AI competitor and where that supply contract is the single biggest piece of disclosed enterprise AI revenue.

    What this IPO most resembles is not Tesla’s 2010 deal or Twitter’s 2013 deal but rather a hybrid of the East India Company chartering and a moonshot R&D vehicle taken public. It is a real cash-flowing business at the Connectivity layer (the largest satellite ISP on Earth) wrapped around a launch monopoly (more than 80% of global mass to orbit) wrapped around a venture-stage AI laboratory (Colossus, Grok, the Anthropic deal, the Cursor option) all underwritten by a CEO compensation structure whose biggest payoffs require a Mars colony. The investor question is not whether any individual piece works, because three of the four pieces clearly do. The question is whether the public market will price the orbital compute and Mars optionality at zero, at a small positive number, or at the eye-watering multiple the $7.5 trillion top tranche of Musk’s pay package implies the board thinks is achievable. There is no precedent for a public company successfully executing on that scale of ambition. There is also no precedent for SpaceX, Starlink, Falcon 9, or Colossus II coming online in 91 days. The S-1 reads like the company assumes the precedent is itself.

    Read the full SpaceX S-1 filing on the SEC EDGAR system for the complete prospectus, including the financial statements and all related disclosures.

  • Gavin Baker on Orbital Compute, TSMC, Frontier AI Models, Anthropic’s Vertical Take Off, and the Coming Wafer Shortage

    Gavin Baker, founder and CIO of Atreides Management, returns to Patrick O’Shaughnessy’s Invest Like the Best for his sixth appearance. He calls the current AI moment the most extraordinary moment in the history of capitalism, walks through what Anthropic’s vertical takeoff in revenue actually means, lays out why orbital compute is closer than skeptics believe, dissects the TSMC bottleneck that may be the only thing standing between today’s market and a full-on AI bubble, and rates every hyperscaler on how they have positioned for a world where frontier model providers may stop selling API access altogether.

    TLDW

    Anthropic added eleven billion dollars of ARR in a single month, which is roughly the combined business of Palantir, Snowflake, and Databricks built over a decade. That is the setup. From there Gavin Baker covers the March and April selloff, the contrarian read that a closed Strait of Hormuz was actually bullish for American manufacturing competitiveness, why Anthropic and OpenAI multiples may be misleadingly cheap on an unconstrained run rate basis, why Elon Musk’s discipline on SpaceX valuation created a superpower of permanent access to capital, the practical engineering case for orbital compute as racks in space rather than Pentagon sized space stations, why TSMC’s capacity discipline is the single most important variable in whether the AI cycle becomes a bubble, what Terafab in Texas changes, why the Pareto frontier of AI models has flipped from Google dominance to Anthropic and OpenAI dominance in nine months, the shift from all you can eat AI subscriptions to usage based pricing and what that means for revenue scaling, Richard Sutton’s bitter lesson as the largest risk to the AI trade, why frontier tokens still capture an overwhelming share of economic value, the role of continual learning as the third great open question, why most new chip startups should not try to build a better GPU, why Cerebras did something different and hard, why disaggregated inference may extend GPU useful lives to ten or fifteen years and rescue the private credit industry, why being in the token path is the new venture filter, the new prisoner’s dilemma around releasing frontier models via API, an honest rating of Google, Meta, Amazon, and Microsoft, why personal safety is becoming a real AI era risk, and why he remains an AI optimist maximalist who believes this could be the next Pax Americana.

    Key Takeaways

    • Anthropic added eleven billion dollars of ARR in one month, more than the combined businesses of Palantir, Snowflake, and Databricks built across a decade. There is no precedent for this in the history of capitalism.
    • The SaaS and cloud revolution created between five and ten trillion dollars of value over twenty years. AI is replaying that compression on a timeline measured in months.
    • The March selloff was a drawdown driven by disagreement with price action, not invalidated thesis. That is the kind of drawdown an investor can lean into.
    • Deep Seek Monday in January 2025 was a similar setup. By the day of the selloff, AWS Asia GPU prices had already doubled, GPU availability had fallen, and it was obvious reasoning models would be vastly more compute hungry at inference. The market priced the opposite.
    • The Strait of Hormuz closing was actually positive for America. US natural gas (the primary input into US electricity, which feeds AI) fell twenty percent on Bloomberg while Asian and European natural gas doubled or tripled. American manufacturing competitiveness improved overnight.
    • The US is now the world’s largest producer and exporter of oil and gas. The economy is dramatically less energy intensive than in the 1970s. The shortage trauma comparison does not hold.
    • Tech as a sector traded as cheaply versus the rest of the market in early April as at any point in the last ten years, into the single most bullish moment for AI fundamentals on record.
    • Anthropic is dramatically more capital efficient than OpenAI, having burned roughly eighty percent less to reach a similar revenue scale. They have very different structural returns on invested capital.
    • Anthropic at roughly nine hundred billion for fifty billion of ARR (growing a thousand percent) is striking. Adjusted for compute constraint, the unconstrained run rate could be one hundred fifty to two hundred billion, putting the implied multiple closer to five times.
    • Claude Opus generates roughly seventy percent fewer tokens for the same question than previously, with token quantity tied to answer quality. Subscribers on flat-fee plans are getting a lobotomized model.
    • Elon Musk’s superpower is twenty years of making investors money. He never pushes valuation. SpaceX compounded low thirty percent per year for a decade because Musk treats fair pricing as a sacred covenant.
    • Capitalism will solve the watts shortage. The current bottleneck has shifted from chips and energy to zoning and political approval. Many capex decisions are paused until after the US midterms.
    • The watts shortage probably begins to alleviate in 2027 and 2028. Orbital compute solves it longer term.
    • Orbital compute is not Pentagon sized data centers in space. It is racks in space. A Blackwell rack is three thousand pounds, eight feet tall, four feet deep, three feet wide. SpaceX has shown a satellite roughly that size.
    • The satellites operate in sun synchronous orbit so solar wings (around five hundred feet per side) always face the sun and the radiator on the dark side always points to deep space.
    • Starlink V3 satellites already run at around twenty kilowatts. A Blackwell rack runs at one hundred kilowatts. SpaceX engineers express genuine confidence they have already solved cooling and radiator design at these scales.
    • Racks in space are connected with lasers traveling through vacuum, the same lasers already on every Starlink. SpaceX operates the world’s largest satellite fleet and, via xAI Colossus, the world’s largest data center on Earth.
    • Inference will move to orbit. Training will stay on Earth for a long time. Terrestrial data centers remain valuable for the rest of an investor’s career.
    • The wafer bottleneck is structural and political. TSMC is essentially Taiwan’s GDP, water, and electricity. The leaders see themselves as inheritors of Morris Chang’s sacred legacy and they do not behave like a Western public company.
    • Jensen Huang has never had a contract with TSMC. The relationship is run on handshakes and the assumption that things will be fair over time.
    • If TSMC did everything Jensen wanted, Nvidia could be selling two to three trillion dollars of GPUs in 2026 and 2027. TSMC’s discipline is the single largest factor preventing a true AI bubble.
    • Historically, foundational technologies always get a bubble. Railroads, canals, the internet. The current AI buildout is overwhelmingly funded out of operating cash flow, GPUs are running at one hundred percent utilization, and that is fundamentally different from the year 2000 fiber overbuild.
    • If one of Intel or Samsung Foundry catches up at the leading node, the other will follow, and TSMC’s discipline collapses. Watch TSMC capacity decisions to predict a bubble.
    • Terafab, the SpaceX and Tesla joint venture to build the world’s largest fab in America, has a partnership with Intel that grants access to fifty years of institutional foundry knowledge. The A teams at ASML, KLA, Lam Research, and Applied Materials will follow Elon’s reputation in hardware engineering.
    • The hiring playbook for Terafab includes building Taiwan Town, Japan Town, and Korea Town next to the fab. Recruit the engineers and import their families, their restaurants, and their staff.
    • Frontier tokens still capture an overwhelming share of all economic value created at the model layer. This is surprising and is one of the three big open questions for AI investing.
    • The Pareto frontier of intelligence versus cost has flipped. Nine months ago Google’s TPU dominated every point on the frontier. Today Anthropic and OpenAI dominate, with Grok 4.3 on the frontier and Gemini 3.1 hanging on.
    • Google’s conservative TPU V8 design (partly an attempt to reduce dependence on Broadcom and Nvidia) is the leading explanation for the loss of per token cost leadership.
    • AI pricing is shifting from all you can eat to usage based, mirroring the cellular and long distance industries. Cellular stopped being a great growth industry when it went all you can eat. AI just made the opposite move.
    • OpenAI and Anthropic together could exceed two hundred billion in ARR this year if compute keeps coming online and frontier token pricing holds.
    • The two hundred fifty dollar a month consumer AI plan is no longer enough to evaluate frontier capability. Enterprise plans with usage based billing are required because rate limits are now severe.
    • The three biggest open questions for AI investors are: violation of the bitter lesson via ASI or human ingenuity, whether frontier tokens keep commanding their premium, and when continual learning arrives.
    • Today’s continual learning is crude reinforcement learning during mid training on verifiable tasks. True continual learning means weights updating dynamically, like a human who learns the first time they touch fire.
    • Trying to build a better GPU is a losing strategy. Jensen will copy any one to three percent share design. Startups should target one percent share, do something different, and make it hard enough that Nvidia cannot fast follow.
    • Disaggregated inference (separating prefill and decode) opens new design canvases. Prefill is memory capacity bound. Decode is memory bandwidth bound. Each can be optimized independently.
    • Cerebras did something different and hard with wafer scale computing. Three generations of chips and real grit to get there.
    • Disaggregation of inference may stretch GPU useful lives to ten or fifteen years, dropping financing costs from low sevens to five or six percent, mathematically lowering the cost of the AI buildout and likely saving the private credit industry from its SaaS loan exposure.
    • Sellers of shortage outperform buyers of shortage. But owning the largest installed base of what is currently in shortage (hyperscaler CPU fleets, for example) is also a strong position.
    • Most of the economic value at the application layer of AI has been destroyed, not created. The exceptions are companies in the token path or in niches small enough that frontier labs ignore them.
    • Coding may be the shortest path to ASI. If you can write code, you can write code that does anything. Cursor, Cognition, and Anthropic correctly focused on it.
    • Jensen could probably get close to the frontier with his own Nemotron family of models whenever he wants. The fact that he chooses not to is a strategic decision about not commoditizing his customers.
    • The new prisoner’s dilemma in AI is whether frontier labs release their best model via API. If everyone agrees not to, Chinese open source falls behind. If anyone defects, the defector pulls ahead on revenue and resources, forcing everyone else to defect.
    • Google still owns the largest compute installed base. Without TPU’s prior cost advantage, this matters more. YouTube data has real value in a world of robotics. GCP is going crazy.
    • Meta deserves credit for becoming AI first internally faster than any other internet giant. Musa, their first MSL model, is impressively close to the Pareto frontier.
    • Amazon is strong because of Trainium and robotics driven retail P&L efficiency. Nova is better than it gets credit for.
    • Microsoft flinched on capex in early 2025 and lost position. Satya Nadella’s current decision to use Microsoft compute for Microsoft products rather than reselling to OpenAI is a courageous and probably correct call, even at the cost of an eight hundred dollar stock price.
    • The hyperscalers most engaged with startups are Amazon and Nvidia by a mile, followed by Google. Broadcom is the favorite ASIC partner. AMD, Microsoft, and Meta have minimal startup engagement and that will cost them as the best teams are now at startups.
    • Personal safety in an AI era requires a family or company safe word that cannot be socially engineered. Deepfake voice and video extortion at the speed of FaceTime is already feasible.
    • Ukraine is winning largely on the back of having the best battlefield AI outside America and Israel. Adversaries are starting to internalize what AI dominance means geopolitically.
    • An optimistic read is that this becomes a new Pax Americana, the way the post 1945 American nuclear monopoly was used to rebuild Germany and Japan rather than dominate.
    • AI cured a friend’s daughter’s rare disease by spinning up a research effort that identified a market drug capable of impacting her condition. That is the upside that keeps Gavin an AI optimist maximalist.

    Detailed Summary

    The most extraordinary moment in the history of capitalism

    Gavin’s framing of the current moment is unusually direct. Anthropic added eleven billion dollars of annual recurring revenue in a single month. The three highest profile SaaS companies of the last decade plus, Palantir, Snowflake, and Databricks, took a decade and tens of thousands of employees collectively to build the combined business that Anthropic added in thirty days. He has been investing through every major tech cycle and says there is no historical analog. Not the dotcom era, not the cloud transition, not mobile. This is its own thing.

    The market response, then, was peculiar. The NASDAQ sold off into the single most bullish moment for AI fundamentals on record. Tech traded at roughly its widest discount versus the rest of the market in a decade. Investors who said they wished they had bought into AI during 2022, during COVID, or during Deep Seek Monday got the same valuation setup again in early April, this time with an even clearer inflection.

    Why the Strait of Hormuz closing was secretly bullish for America

    One reason the macro fear in March may have been mispriced is that the same geopolitical event that drove the selloff was, in practice, a relative benefit to the United States. American natural gas, the input into American electricity, which is the input into American AI training and inference, fell roughly twenty percent. Asian and European natural gas prices doubled or tripled. The US emerged with sharply improved relative manufacturing competitiveness, which is exactly what the current administration cares about.

    The 1970s comparison does not hold. The US economy is dramatically less energy intensive, it is now the world’s largest producer and largest exporter of oil and gas, and there are no shortages, only price moves. That backdrop made it easier for disciplined investors to stay focused on AI fundamentals through the volatility.

    Anthropic and OpenAI valuations on an unconstrained run rate

    Anthropic at roughly nine hundred billion for fifty billion of ARR sounds rich until you adjust for the fact that the company is severely compute constrained. Gavin estimates that, unconstrained, Anthropic might be at one hundred fifty to two hundred billion in run rate revenue, putting the implied multiple closer to five times. He also points out that Claude Opus now generates roughly seventy percent fewer tokens for the same question than it used to. Token quantity correlates with answer quality, and Anthropic is rate limiting and shrinking outputs to ration capacity across its user base.

    Anthropic and OpenAI are also structurally very different. Anthropic has burned around eighty percent less cash than OpenAI to reach a comparable revenue scale. That implies very different long term returns on invested capital, though OpenAI has done a better job locking in compute and Sarah Friar is one of the most exceptional CFOs Gavin has worked with.

    Why neither lab is raising at a three trillion dollar valuation

    The answer Gavin gives is that both labs are deliberately leaving valuation on the table the way Elon has done for two decades. SpaceX compounded at low thirty percent annually for a decade because Elon never pushed price. The result is a permanent superpower of access to capital. Investors trust him because they have made money with him for twenty years. That is a moat that compounds with every round.

    Anthropic could probably raise at a one hundred percent premium to its rumored latest mark. They are choosing not to. In an uncertain world (Ukraine, Russia, Iran, Taiwan), preserving the ability to raise more capital later at fair prices is more valuable than maximizing this round.

    Watts and wafers, the two real constraints

    Capitalism is solving the watts problem. The leading PE infrastructure investors now say zoning and political approval, not chips or energy, are the gating factors. Companies are deferring big capex announcements until after the US midterms. Turbine capacity is being doubled at the manufacturers. Companies like Boom Aerospace are repurposing jet engines for grid use. Watts probably ease meaningfully in 2027 and 2028 and then orbital compute does the rest.

    Wafers are the harder problem because they live in Taiwan, run on handshakes, and depend on a corporate culture that does not respond to public market incentives. TSMC is essentially the GDP, water consumption, and electricity consumption of Taiwan. Its leadership treats the company as the legacy of Morris Chang. The Silicon Shield doctrine is real and internal.

    Orbital compute as racks in space

    The biggest mental update Gavin asks listeners to make is to stop picturing data centers in space as Pentagon sized space stations. A Blackwell rack is three thousand pounds and roughly the size of a refrigerator. SpaceX has shown a concept satellite of about that size. Solar wings extend five hundred feet to each side and the radiator extends hundreds of feet behind, both possible because the orbit is sun synchronous and the orientation is fixed relative to the sun.

    SpaceX engineers Gavin has spoken to at Starbase express genuine confidence that they have solved cooling at these power levels. They have. Starlink V3 satellites already operate at twenty kilowatts. A Blackwell rack is one hundred kilowatts. The same company operates the world’s largest satellite fleet and the world’s largest data center on Earth via xAI Colossus. The racks are connected to each other with lasers traveling through vacuum, technology already deployed in every Starlink. The naysayers, Gavin observes, are armchair skeptics and Larry Ellison’s response (he is out there landing rockets, no one else is) is the right frame.

    Terafab in Texas and the threat to TSMC’s discipline

    Terafab, the SpaceX and Tesla joint venture, intends to be the largest fab in the world. The partnership with Intel grants access to fifty years of foundry institutional knowledge, allowing Terafab to start three to five quarters behind the leading node rather than fifteen years behind. The A teams at the semicap equipment companies (ASML, KLA, Lam Research, Applied Materials) will follow Elon’s reputation in hardware engineering the same way they followed TSMC twenty years ago when Intel stumbled.

    The talent strategy is the part most observers underestimate. Recruit the best engineers globally, then import their families, their restaurants, their staff. Build Taiwan Town, Japan Town, and Korea Town next to the fab. Optimize the human experience for the people whose work matters. Intel and Samsung do not think that way.

    Bubble watch and the year 2000 comparison

    Every foundational technology in modern history has had a bubble. Railroads, canals, the internet. Carlota Perez documented why. Markets correctly identify the importance, diversity of opinion collapses, supply gets ahead of demand, the bubble crashes. The current cycle has two important differences. The buildout is overwhelmingly funded out of operating cash flow, not debt. Every GPU is running at one hundred percent utilization, while at the peak of the fiber bubble ninety nine percent of fiber was unused.

    TSMC discipline is the single largest reason a bubble has not formed. If Jensen could buy everything TSMC could theoretically make, Nvidia could sell two to three trillion dollars of GPUs in 2026 and 2027. At some point that becomes more than the market can absorb. If Intel or Samsung Foundry catches up at the leading node, the other will too. TSMC’s pricing discipline collapses and the bubble starts.

    The Pareto frontier and the loss of Google’s cost advantage

    The most important chart in AI is the Pareto frontier of model intelligence versus per token cost. Nine months ago, Google’s TPU based models dominated every point on it. OpenAI, Anthropic, and xAI sat inside the frontier. Today the frontier is dominated by Anthropic and OpenAI, with Grok 4.3 on the frontier and Gemini 3.1 hanging on by subsidization more than economics. The most likely cause is Google’s conservative TPU V8 design, an attempt to reduce dependence on Broadcom and Nvidia that sacrificed per token economics.

    The bitter lesson, frontier tokens, and continual learning

    Three open questions dominate AI investing. The first is whether Richard Sutton’s bitter lesson (more compute beats human algorithmic cleverness) gets violated by ASI itself optimizing for efficiency. Closer observers of AI are more skeptical of a violation. Gavin thinks ASI’s first move will be to make itself more efficient and more resourced, which is technically a temporary violation.

    The second is whether frontier tokens keep capturing the overwhelming share of economic value at the model layer. Today they do, surprisingly. Gemini 3.1 Pro was mindblowing nine months ago and is intolerable today. The third is when continual learning arrives. Today’s models need a million fire touches to learn what a human learns from one. True continual learning would mean dynamic weight updates in real time and would produce a fast takeoff.

    From all you can eat to usage based AI pricing

    AI is shifting from flat fee plans to usage based pricing. The historical analogy is cellular and long distance. Both stopped being great growth industries when they went all you can eat. AI just made the opposite move. The consequence is that flat fee subscribers, even on premium consumer plans, get a rate limited and token throttled version of the frontier model. Enterprise plans with usage based billing are now required to evaluate true capability. Gavin thinks the combination of new compute coming online and usage based pricing is what gets OpenAI and Anthropic past two hundred billion in combined ARR this year.

    Chip startups, prefill decode disaggregation, and Cerebras

    Trying to build a better GPU is the wrong move. The four scaled players (Nvidia, AMD, Trainium, TPU) have copy capability for any one to three percent share design that looks attractive. The good news for startups is that disaggregated inference (separating prefill and decode) opens a richer design canvas. Prefill is memory capacity bound. Decode is memory bandwidth bound. Each can be optimized independently. Andrew Fox’s analogy is a British naval ship of the eighteenth century. Prefill is loading the cannon. Decode is firing it.

    Cerebras is the model. Wafer scale computing is genuinely different and genuinely hard. It took three generations of chips to get right. Andrew Feldman and his team had the grit to keep going through chip one being a failure. The design has a high ratio of on chip compute and memory relative to shoreline IO, which is why Cerebras is now experimenting with putting an optical wafer on top of the compute wafer to solve scale out.

    GPU useful lives and the rescue of private credit

    One of the strongest claims in the conversation is that disaggregated inference will stretch GPU useful lives to ten or fifteen years. The skeptical narrative (GPUs are obsolete in two years, companies are cooking their depreciation books) is wrong. You can put a Cerebras system or Groq LPU in front of older Hopper or Ampere parts, use them only for prefill, and run them until they physically melt. Private credit, which is in pain from SaaS loans and which underwrote GPU loans on three to four year lives, may be saved by this.

    If GPU financing rates can come down from low sevens to five or six percent, the mathematics of the AI buildout improves materially. That is a structural tailwind that compounds for years.

    The application layer, the token path, and a new prisoner’s dilemma

    Trillions of dollars of value have been destroyed at the application layer, not created. Cursor and Cognition are the rare scaled exceptions, and they got there by focusing on coding very early. As Amjad Masad noted, coding is plausibly the shortest path to ASI because a coding agent can write itself into any new domain. Jamin Ball’s frame is that the new venture filter is whether the company is in the token path. Data Bricks is. Most application layer startups are not.

    Jensen could probably get close to the frontier with Nemotron whenever he wants, and the strategic question of whether to do that is a new prisoner’s dilemma. If every frontier lab agrees not to release best models via API, Chinese open source falls steadily behind. If anyone defects, the defector gains revenue and resources, and everyone else has to defect. The same dynamic exists between TSMC, Intel, and Samsung. If Nvidia or AMD ever truly used an alternative foundry, that foundry would catch up rapidly.

    Rating the hyperscalers

    Google has the largest compute installed base, the YouTube data that matters in a robotics world, and a search business that prints. Their loss of TPU cost leadership is the surprise of the year. If Google IO in five days does not produce a leapfrog model, the Nvidia centric narrative gets even stronger.

    Meta deserves real credit. Zuckerberg made Meta AI first internally faster than any other internet giant, paid up for the talent contracts when no one else would, and shipped Musa as a first model from MSL that is close to the Pareto frontier. Amazon is well positioned on Trainium, robotics in retail, and a Nova model line that is better than it gets credit for. Microsoft flinched on capex in early 2025 and lost position. Satya Nadella’s current decision to use Microsoft compute for Copilot rather than reselling to OpenAI is courageous and probably correct, even at the cost of stock price.

    The most interesting cross hyperscaler metric is startup engagement. Nvidia and Amazon engage deeply with startups. Google is next. Broadcom is the favored ASIC partner. AMD, Microsoft, and Meta have minimal startup engagement, which Gavin believes will cost them as the best teams now sit at startups.

    Personal safety, geopolitics, and the Pax Americana case

    The closing section turns darker. Personal safety in an AI era requires a family or company safe word that cannot be socially engineered. Deepfake voice and video extortion via something that looks exactly like your child calling on FaceTime is already feasible. Political violence against AI leaders is a real concern. Geopolitically, Ukraine is winning largely because it has the best battlefield AI outside America and Israel. How adversaries respond to that asymmetry is the next great variable.

    Gavin’s optimistic frame is the Pax Americana. After 1945 the US had a nuclear monopoly and could have controlled the world. Instead it rebuilt Germany and Japan, both of which became the most reliable American allies for the next eighty years. If AI dominance plays out similarly, this is a generationally positive story rather than a destabilizing one. The personal anecdote that closes the conversation is a friend whose daughter was diagnosed with a rare genetic condition. He spun up agents, identified a drug already on the market that addresses her mutation, and her life is immeasurably different because of AI. That is the upside.

    Thoughts

    The Anthropic eleven billion in a month framing is the kind of stat that resets priors. The right way to interpret it is not as a one off but as a measure of how fast value can compound when the underlying technology improves on a curve steeper than the ability of the rest of the economy to absorb it. The skeptical question is whether that ARR is durable or whether it is heavily tied to a customer base of other AI companies that are themselves on a single venture funded year of runway. The bullish answer is that frontier coding, frontier research, and frontier enterprise tasks are not going to stop being valuable, and Anthropic is the best at all three. Both can be true. The number is still extraordinary.

    The argument that TSMC discipline is the only thing preventing a bubble is the analytically tightest part of the conversation. The implied trade is to watch TSMC capacity additions like a hawk and to be more, not less, cautious if Intel Foundry or Samsung Foundry ever announce real share at the leading node. The Terafab thesis is more speculative but more interesting. If Elon’s talent recruiting playbook works and the Intel partnership gives Terafab a real seat at the table within five years, the geometry of the global semiconductor industry shifts in a way that is bullish for American manufacturing, bullish for power and water infrastructure in Texas, and ambiguous for TSMC itself.

    The Pareto frontier discussion deserves more attention than it usually gets. Pricing leadership in AI is not a vanity metric. It determines who can subsidize free tier usage, who can absorb compute shortages, who can ship cheaper enterprise plans, and ultimately whose model becomes the default for any given workload. Google losing per token leadership in nine months is one of the most under analyzed events in the sector and it explains a lot about why Anthropic and OpenAI are growing the way they are. If Google IO does not produce a leapfrog model, the implied verdict on TPU V8 design choices gets a lot harsher.

    The application layer destruction point is worth sitting with. Founders building on top of frontier models are competing in a world where the model itself moves faster than any moat they can build, where the model lab can absorb their niche if it gets interesting, and where the only protection is either deep token path integration or a niche so small the lab does not bother. That is a much harsher venture environment than the early SaaS era. The compensating opportunity is that one human can now run a hundred agents, so the ceiling on what a small team can build is correspondingly higher. The bet is that productivity per founder rises faster than competitive pressure from the labs. We will find out.

    The orbital compute pitch is the section that will polarize listeners. The naive read is that this is science fiction. The closer read is that every component (sun synchronous orbit, laser interconnect, twenty kilowatt satellite buses, ten thousand satellite manufacturing cadence, full rocket reusability) already exists. The remaining engineering problems are repair, maintenance, and radiator scale, all of which are real but tractable on a five to ten year horizon. The strategic implication is that the political and zoning ceiling on terrestrial data centers becomes less binding if orbital compute is a credible alternative for inference workloads. The investor implication is that being short the watts and cooling complex on a five year horizon is a real trade, not a meme.

    Watch the full conversation here.

  • How the X For You Algorithm Works (May 15, 2026 Source Code Walkthrough)

    Diagram of the X For You algorithm pipeline showing content signals and user signals flowing through embed, attend, score, rank, and filter stages into a curated feed

    On May 15, 2026, xAI shipped a major update to the open-source release of the X “For You” recommendation algorithm. The repository now includes a runnable end-to-end inference pipeline, a pre-trained mini Phoenix transformer, a brand-new content-understanding service called Grox, and ad-blending logic. This is the most transparent look at how a major social feed actually picks your posts that has ever been published.

    This is the practical, plain-english guide. We read the source. Here is exactly how a post travels from your fingertips to someone’s For You tab, and what you can do to be the post that wins.

    The whole strategy in one sentence

    Write posts people reply to, repost, DM to a friend, linger on, and follow you for. Avoid anything that earns a mute, block, report, or spam flag. Space your posts hours apart. That is the algorithm.

    TL;DR

    • The For You feed is no longer a stack of heuristics. It is a single transformer-based machine learning system that predicts the probability you will like, reply, repost, share, dwell on, or hide a given post.
    • Posts come from two pools: Thunder (people you follow) and Phoenix Retrieval (the rest of X, found by similarity search).
    • A model called Phoenix scores every candidate against your engagement history. The final score is a weighted sum of nineteen predicted actions, with negative weights for “block”, “mute”, and “report.”
    • Almost no hand-engineered features survive. Freshness, verification badges, follower counts, and post type are not directly boosted. They are signals the transformer learns to use from your behaviour.
    • Out-of-network content is penalised by a tunable factor, so in-network posts have an edge by default.
    • A separate service called Grox continuously classifies posts for spam, policy violations, and brand safety. Flagged content gets filtered or de-amplified before it reaches scoring.
    • The best optimisations are still the boring ones: write posts that earn long dwell time, replies, reposts, and follows, and avoid anything that triggers mutes or reports.

    What changed on May 15, 2026

    The January 2026 release gave us the architecture but not a working system. The May update is the one that matters:

    • phoenix/run_pipeline.py replaces the separate retrieval and ranking scripts with a single inference entry point that mirrors production.
    • A pre-trained mini Phoenix model (256-dim embeddings, 4 attention heads, 2 transformer layers) is bundled as a roughly 3 GB Git LFS archive. You can run inference without training.
    • The Grox content-understanding service is now public. It runs classifiers and embedders for spam detection, post categorisation, and policy enforcement.
    • Ads blending is now in the open. So is brand-safety tracking.
    • New query hydrators mean the model sees your followed topics, starter packs, served history, impression bloom filters, IP, and mutual-follow graph at request time.
    • New candidate hydrators add engagement counts, language codes, media detection, quote post expansion, and mutual follow scores.
    • New candidate sources for ads, who-to-follow, Phoenix Mixture-of-Experts, Phoenix Topics, and prompts.

    The For You pipeline in one picture

    Every time you pull to refresh, the same pipeline runs:

    1. Query hydration: pull the requesting user’s engagement history, follow list, topics, served history, and metadata.
    2. Candidate sourcing: gather candidates from Thunder and Phoenix Retrieval in parallel.
    3. Candidate hydration: enrich each candidate with text, media, author, engagement counts, brand-safety labels, language, mutual follow scores.
    4. Pre-scoring filters: drop duplicates, posts that are too old, your own posts, blocked or muted authors, posts you’ve already seen, and posts with your muted keywords.
    5. Scoring: run candidates through Phoenix, combine the predicted action probabilities into a weighted score, attenuate repeated authors, penalise out-of-network content.
    6. Selection: sort by score, take the top K.
    7. Post-selection filtering: final visibility check for deleted, spam, violence, gore, abuse, and de-duplication of conversation branches.
    8. Side effects: cache request info, then return the ranked feed.

    That is the whole story. Every choice the system makes lives in one of those stages.

    Where candidates come from: Thunder vs Phoenix

    Two sources feed the pipeline.

    Thunder is the in-network store. It is an in-memory firehose that consumes post create and delete events from Kafka and serves sub-millisecond lookups for recent posts from people you follow. Posts older than the retention window get trimmed automatically. Thunder is why the feed feels fast.

    Phoenix Retrieval is the out-of-network source. It is a classic two-tower neural network. A user tower encodes your features and recent engagement history into a single embedding. A candidate tower does the same for every post in the global corpus. The system then does an approximate nearest-neighbour search over those embeddings to find the posts whose vectors point most similarly to yours. A dot product between vectors is all it takes.

    How many candidates from each side? The code does not hard-code a ratio. It is set at runtime via parameters (ThunderMaxResults and PhoenixMaxResults). New users get a different retrieval cluster while their account is below an age threshold and a minimum-following count.

    The creator implication is the part most guides miss: follower count is not what gets you into out-of-network feeds. Embedding similarity is. Phoenix knows nothing about how famous you are. It knows that the people who engage with posts like yours have engagement histories that look like the histories of users it is trying to serve.

    How Phoenix ranks posts

    After candidates arrive, every one of them gets a score from the Phoenix ranking transformer. The architecture, per phoenix/README.md, is small by language-model standards:

    • 128-dimensional embeddings
    • 4 transformer layers
    • 4 attention heads
    • 127-position user history sequence
    • 64-position candidate sequence
    • 1,000,000 entries each in the user, item, and author vocabularies (with 2 hash functions per entity)
    • 19 predicted action types

    The transformer’s input is a sequence of your past engagements. Each engagement carries the post you engaged with, the author, the action you took, and the product surface (For You, profile, search). The candidates are appended as a second segment. The model uses candidate isolation masking: candidates can attend to your history but not to each other. This is a deliberate engineering choice. It means a post’s score does not depend on the other posts in the batch, which keeps scoring cacheable and consistent.

    The output is one probability per action type, per candidate.

    The action weights: what positive and negative engagement is worth

    The Weighted Scorer combines those probabilities into a single number:

    “ Final Score = Σ (weight_i × P(action_i)) “

    The exact weight values are not in the open-source repo. They live in an external configuration crate (xai_home_mixer) that xAI tunes continuously. What the repo does show us is the shape of the signal, and that is what matters for strategy.

    Positive weights are applied to these predicted actions:

    • favorite
    • reply
    • retweet
    • quote
    • quoted click
    • click
    • profile click
    • photo expand
    • video view (only counted if the video is above a minimum duration)
    • share
    • share via DM
    • share via copy link
    • dwell (the user lingered on the post)
    • continuous dwell time (how long they lingered)
    • follow author

    Negative weights are applied to:

    • not interested
    • block author
    • mute author
    • report

    A few observations matter for creators. Replies, reposts, follows, and DM shares are listed as their own separately-weighted actions, which is the strongest signal we have that xAI treats them as more valuable than a like. Dwell and continuous dwell time are split into two predictions, which means how long the average person reads your post is its own ranking lever. And the negative actions are not just filters. They actively push the score down for posts that even slightly resemble content that triggers mutes or reports.

    The out-of-network penalty

    Out-of-network content does not start on an even footing with in-network content. The OON Scorer multiplies a candidate’s score by a configurable factor that is less than one when the candidate came from Phoenix Retrieval. This is why following the right accounts still matters.

    There is an explicit override for new users: if your account is fresh and you follow at least the minimum number of accounts, the OON penalty is softened with a more permissive factor. This is the bootstrap mechanism that gets new users a populated For You feed before they have generated enough engagement history for Phoenix to personalise around.

    The author diversity damper

    Once posts are sorted, the Author Diversity Scorer runs. It applies an exponential attenuation to repeated authors:

    “ multiplier(position) = (1 - floor) × decay^position + floor “

    The first post from an author keeps its full score. The second is attenuated. The third more so. A floor value prevents an author from being attenuated below a minimum. The decay and floor are tunable parameters, not constants. The practical consequence is that posting fifteen times in twenty minutes does not produce fifteen top-of-feed impressions. It produces one or two.

    What Grox does (and why it is the most important new piece)

    The grox/ directory is new in May 2026 and is the single most underreported part of the update. It is an asynchronous task-execution engine that runs content classifiers and embedders on every post. Among the tasks present in the repo:

    • Spam detection, including a low-follower reply-spam classifier
    • Safety policy classification across categories like violent media, adult content, hate, self-harm, and platform-policy violations
    • A “post safety screen deluxe” pipeline that re-checks adult content classification with a second pass
    • Media classification for images and video
    • Multimodal post embedding for retrieval and ranking

    Grox does not directly score posts. It produces labels. Those labels are consumed by the visibility filter (VFFilter) at the post-selection stage and by the brand-safety hydrator that ad placement uses. The effect is that posts the system thinks are spam, policy-violating, or unsafe for ads next to do not get removed entirely. They get de-amplified before they ever reach the Phoenix ranking step, or filtered after.

    If you have wondered why a perfectly reasonable post sometimes mysteriously underperforms, this is the likely culprit. A Grox classifier flagged something.

    What the algorithm explicitly does not boost

    This is worth saying twice because the rumour mill keeps repeating the opposite. In the open source code:

    • There is no verified-badge boost in the scorer.
    • There is no follower-count boost.
    • There is no link penalty. Links are not separately weighted.
    • There is no freshness boost. The age filter removes posts above a threshold but does not score newer posts higher.
    • Subscriber status is used for filtering paywalled content, not for boosting reach.

    If verification, Premium, or any of these correlate with reach in practice, it is because the Phoenix transformer has learned to predict that users engage with those posts more, not because a hand-written rule said so. The whole point of the architecture, per the repo, is that every such heuristic has been removed and the model learns the signal from your engagement sequences.

    How ads get inserted

    Ads ride along through the same pipeline. The blender requires at least five organic posts before an ad can be placed. It computes a spacing interval, partitions candidate ads by brand-safety verdict, and caps the number of ads from the safe set to roughly half the safe-set size. A second layer of contextual checks drops ads when neighbouring posts have a weak brand-safety rating, a conflicting handle, or a keyword collision. The result is an interleaved feed that tries to keep brand-safety risk down without starving the auction.

    How creators should post for the 2026 algorithm

    Stop optimising for proxies. Optimise for the actions Phoenix is actually predicting:

    1. Write for replies and reposts, not for likes. A like is one positive weight. A reply, a repost, a quote, and a follow are each separately weighted on top. Posts that ask a question, take a stance, or offer a frame for someone else to argue with consistently outperform posts that close a thought.
    2. Aim for dwell. A long-form thread, a clear photo, or a video that people watch to the end gets two positive signals: dwell and continuous dwell time. A one-line post you scroll past in a quarter-second gets neither.
    3. Earn the follow. Follow-author is a predicted action with its own weight. A post that successfully sells a new viewer on hitting follow scores more than a post that doesn’t.
    4. Do not cluster. Author Diversity attenuates your second and third posts inside the same scoring window. If you have three things to say, space them out by hours, not minutes.
    5. Avoid anything that gets you muted, blocked, or reported. Those carry explicit negative weights. Engagement bait that produces a single block does measurable damage to the score of that post and any signal it sends about the author.
    6. Do not be Grox-flagged. Spam-shaped behaviour (reply-bombing with the same line, posting at high frequency with a low follower count, low-quality media) gets you classified by Grox before you ever reach the scorer.
    7. Follow more accounts in your niche. Phoenix Retrieval is similarity-based, but the OON penalty means in-network candidates still have a head start. The denser your in-network graph in your niche, the more likely your posts surface there.
    8. Build an engagement history that Phoenix can recognise. The user tower encodes your recent engagement. If you want your content to surface to people who like topic X, engage like a person who likes topic X. The model will learn to send your posts to that cluster.
    9. Lean into video and photo. Photo expand and video view are both separately weighted positive actions. They give a single post more ways to score.
    10. DM-share-worthiness is a quiet superpower. Share via DM and share via copy link are each their own weighted action. A post worth sending to a specific friend is, mechanically, a higher-scoring post than a post merely worth liking.

    Can I run the algorithm locally?

    Yes. With the May 15 release, the runnable inference path is phoenix/run_pipeline.py, and the bundled mini Phoenix checkpoint is enough to score sample posts. You can clone the repo, pull the LFS archive, and watch the pipeline rank a batch end to end. This is, as far as we know, the first time a production-scale social recommendation system has shipped a runnable inference path to the public.

    What is next

    Two trends are worth watching. The first is the cadence: xAI has been pushing material updates every few weeks. Expect the action weights, retrieval ratios, and Grox classifier set to keep moving. The second is the architecture: candidate sources for “Phoenix MoE” and “Phoenix Topics” suggest the next direction is multiple specialised ranking experts rather than a single transformer, with topic awareness fed in explicitly. Promptable feeds (telling X in natural language what you want more of) are the user-visible end of that trend.

    The closing point is the practical one. The 2026 For You algorithm is, more than any version before it, a measurement of how people respond to your post. Strategies that try to game routing, freshness, or format are landing in a system that does not care about those things directly. Strategies that earn replies, holds, follows, and shares are landing in a system that is built, end to end, to reward exactly that.

  • Alex Wang on Leaving Scale to Run Meta Superintelligence Labs, MuseSpark, Personal Super Intelligence, and Building an Economy of Agents

    Alex Wang, head of Meta Superintelligence Labs, sits down with Ashley Vance and Kylie Robinson on the Core Memory podcast for his first long-form interview since Meta’s quasi-acquisition of Scale AI roughly ten months ago. He walks through how MSL is structured, why Llama was off-trajectory, what made MuseSpark’s token efficiency surprise the team, how Meta thinks about a future “economy of agents in a data center,” and where he lands on safety, open source, robotics, brain computer interfaces, and even model welfare.

    TLDW

    Wang explains that Meta Superintelligence Labs is a fully rebuilt frontier effort organized around four principles (take superintelligence seriously, technical voices loudest, scientific rigor, big bets) and three velocity levers (high compute per researcher, extreme talent density, ambitious research bets). He confirms Llama was off the frontier when he arrived, so MSL rebuilt the pre-training, reinforcement learning, and data stacks from scratch. MuseSpark is described as the “appetizer” on the scaling ladder, notable for its strong token efficiency, with much larger and stronger models coming in the coming months. He pushes back on the mercenary narrative around recruiting, frames Meta’s edge as compute plus billions of consumers and hundreds of millions of small businesses, sketches a vision of personal super intelligence delivered through Ray-Ban Meta glasses and WhatsApp, and outlines why physical intelligence, robotics (the new Assured Robot Intelligence acquisition), health super intelligence with CZI, brain computer interfaces, and even model welfare are core to Meta’s roadmap. He dismisses reported infighting with Bosworth and Cox as gossip, declines to comment on the Manus situation, and says safety guardrails (bio, cyber, loss of control) are why MuseSpark cannot currently be open sourced, while smaller open variants are being prepared.

    Key Takeaways

    • Meta Superintelligence Labs (MSL) is the umbrella, with TBD Lab as the large-model research unit reporting directly to Alex Wang, PAR (Product and Applied Research) under Nat Friedman, FAIR for exploratory science, and Meta Compute under Daniel Gross handling long-term GPU and data center planning.
    • Wang says Llama was not on a frontier trajectory when he arrived, so MSL had to do a “full renovation” of the pre-training stack, RL stack, data pipeline, and research science.
    • The first cultural fix was getting the lab to “take superintelligence seriously” as a near-term, achievable goal, not an abstract bet. Big incumbents often lack that religious conviction.
    • Four MSL principles: take superintelligence seriously, let technical voices be loudest, demand scientific rigor on basics, and make big bets.
    • Three velocity levers Wang identified for catching and overtaking the frontier: high compute per researcher, very high talent density in a small team, and willingness to fund ambitious research bets.
    • Wang rejects the mercenary recruiting narrative. He says most hires had strong financial prospects at their prior labs already and joined for compute access, talent density, and the chance to build from scratch.
    • On the famous soup story, Wang neither confirms nor denies Zuck personally made the soup, but says recruiting was highly individualized and signaled how seriously Meta cared about each researcher’s agenda.
    • Yann LeCun publicly called Wang young and inexperienced. Wang says they reconciled in person at a conference in India where LeCun congratulated him on MuseSpark.
    • Sam Altman, asked by Vance for comment, “did not have flattering things to say” about Wang. Wang hopes industry animosities subside as systems approach superintelligence.
    • Wang’s management philosophy borrows the Steve Jobs line: hire brilliant people so they tell you what to do, not the other way around.
    • MuseSpark is framed as an “appetizer” data point on the MSL scaling ladder, not a flagship.
    • The MuseSpark program is built around predictable scaling on multiple axes: pre-training, reinforcement learning, test-time compute, and multi-agent collaboration (the 16-agent content planning mode).
    • MuseSpark outperformed internal expectations and showed emergent capabilities in agentic visual coding, including generating websites and games from prompts, helped by combined agentic and multimodal strength.
    • MuseSpark’s biggest external signal is token efficiency. On benchmarks like Artificial Analysis it hits similar results with far fewer tokens than competitor models, which Wang attributes to a clean stack rebuilt by experts rather than inefficiencies patched by longer thinking.
    • Larger MSL models are arriving in the coming months and Wang expects them to be state of the art in the areas MSL is focused on.
    • The Meta strategic edge: massive compute, billions of consumers across the family of apps, and hundreds of millions of small businesses already on Facebook, Instagram, and WhatsApp.
    • Wang’s headline framing: Dario Amodei talks about a “country of geniuses in a data center.” Meta is targeting an “economy of agents in a data center,” with consumer agents and business agents transacting and collaborating.
    • Consumer AI sentiment is in the toilet because, unlike developers who have had a Claude Code moment, ordinary people have not yet experienced AI as a genuine personal agency unlock.
    • Wang acknowledges the product overhang. Meta held back from deep AI integration across its apps until the models were good enough, and is now entering the integration phase.
    • Ray-Ban Meta glasses are the canonical example of personal super intelligence hardware, with the model seeing what the user sees, hearing what they hear, capturing context, and surfacing proactive insights.
    • Wang admits even AI-native users like Kylie Robinson, who lives in WhatsApp, have not naturally used Meta AI yet. He bets that better models plus deeper integration close that gap.
    • On the competitive landscape: a year ago everyone assumed ChatGPT had already won consumer. Claude Code has since become the fastest growing business in history, and Gemini has taken consumer market share. Wang’s read: AI is far from endgame and each new capability tier unlocks a new dominant form factor.
    • On open source: MuseSpark triggered guardrails in Meta’s Advanced AI Scaling Framework around bio, chem, cyber, and loss-of-control risks, so it is not currently safe to open source. Smaller, derived open variants are actively in development.
    • Meta remains committed to open sourcing models when safety allows, drawing a line through the Open Compute Project legacy and Sun Microsystems open-software heritage.
    • Wang dismisses reporting about a Wang-Zuck versus Bosworth-Cox split as “the line between gossip and reporting is remarkably thin.” He says leadership is aligned on needing best-in-class models and product integration.
    • On the Manus situation, Wang says it is too complicated to discuss publicly and that the deal status implies “machinations are still at play.”
    • On China, Wang separates the people from the state. He still wants to work with talented Chinese-born researchers regardless of his views on the Chinese Communist Party and PLA, which he sees as taking AI extremely seriously for national security.
    • The full-page New York Times AI war ad Wang ran while at Scale was meant to push the US government to treat AI as a step change for national security. He thinks events since then, including DeepSeek and other shocks, have proved that plea correct.
    • On Anthropic’s doom posture, Wang largely agrees with the core message that models are already very powerful and getting more so, while declining to endorse every specific claim.
    • Meta has acquired Assured Robot Intelligence (ARRI), an AI software company building models for hardware platforms, not a hardware maker itself.
    • Wang frames physical super intelligence as the natural sequel to digital super intelligence. Robotics, world models, and physical intelligence all benefit from the same scaling that drives language models.
    • On health, MSL is building a “health super intelligence” effort and will collaborate closely with CZI. Wang sees equal global access to powerful health AI as a uniquely Meta-shaped delivery problem.
    • Wang admires John Carmack but says nobody really knows what Carmack is currently working on. No band reunion announced.
    • The mango model is “alive and kicking” despite rumors. Wang notes MSL gets a small fraction of the rumor-mill attention other labs get and feels sympathy for them.
    • On model welfare, Wang says it is a serious topic that “nobody is talking about enough” given how integrated models have become as work partners. He references research, including from Eleos, that measures subjective experience of models.
    • Wang’s critical-path technology list: super intelligence, robotics, brain computer interfaces. The infinite-scale primitives behind them are energy, compute, and robots.
    • FAIR’s brain research program Tribe hit a milestone called Tribe B2: a foundation model that can predict how an unknown person’s brain would respond to images, video, and audio with reasonable zero-shot generalization.
    • Wang’s main philosophical break with Elon Musk: research itself is the primary activity. Building super intelligence is a research expedition through fog of war, and sequencing of bets really matters.
    • Personal notes: Wang moved from San Francisco to the South Bay, treats Palo Alto as his city now, was a math olympiad competitor, says his favorite activities are reading sci-fi and walking in the woods, and bonds with Vance over country music.

    Detailed Summary

    How MSL Is Actually Organized

    Meta Superintelligence Labs sits as the umbrella organization that Wang oversees. Inside it, TBD Lab is the large-model research group where the most discussed researchers and infrastructure engineers sit, and they technically report to Wang. PAR, Product and Applied Research, is led by Nat Friedman and owns deployment and product surfaces. FAIR continues to run exploratory science, including work on brain prediction models and a universal model for atoms used in computational chemistry. Sitting alongside MSL is Meta Compute, run by Daniel Gross, which owns the long-horizon GPU and data center plan that everything else relies on. Chief scientist Shengjia Zhao orchestrates the scientific agenda across the whole lab.

    Why Wang Left Scale

    Wang says progress in frontier AI has been faster than even insiders expected. Two structural beliefs pushed him toward Meta. First, the labs that actually train the frontier models are accruing disproportionate economic and product rights in the AI ecosystem. Second, compute is the dominant scarce input of the next phase, so the right mental model is to treat tech companies with compute as fundamentally different animals from companies without it. Meta has both, Zuck is “AGI pilled,” and the personal super intelligence memo Zuck published roughly a year ago became the shared north star.

    The Diagnosis: Llama Was Off-Trajectory

    When Wang arrived, the existing AI org needed a reset because Llama was not on the same trajectory as the frontier. The plan he laid out has four cultural principles. Take superintelligence seriously as a real near-term target. Make technical voices the loudest in the room. Demand scientific rigor and focus on basics. Make big bets. On top of that, three structural levers were used to set velocity. Push compute per researcher much higher than at larger labs where compute is diluted across too many efforts. Keep the team small and extremely cracked. Allocate a meaningful share of resources to ambitious, paradigm-shifting research bets rather than incremental refinement.

    Recruiting, Soup, and the Mercenary Narrative

    Wang argues the reporting on MSL hiring overstated the money story. Most of the people MSL recruited had strong financial paths at their previous employers, so individualized recruiting was more about computing access, talent density, and the ability to make big research bets. The recruitment blitz happened fast because Wang knew the team needed to exist “yesterday.” Asked about Mark Chen’s claim that Zuck made soup to recruit people, Wang refuses to confirm or deny who made it but agrees the process was intense and personal. Visitors from other labs reportedly tell Wang the MSL culture feels like early OpenAI or early Anthropic, which lands as the strongest endorsement he could ask for.

    Receiving the Public Hits: Young, Inexperienced, Mercenary

    LeCun called Wang young and inexperienced shortly after departing. The two reconnected in India a few weeks later and LeCun congratulated Wang on MuseSpark. Wang says the age critique has followed him since his earliest Silicon Valley days, so he barely registers it. Altman, asked off-camera by Vance about Wang’s appearance on the show, had nothing flattering to add. Wang’s response is to bet that as the field gets closer to actual super intelligence, the personal animosities will subside. Whether they will is, as Vance puts it, an open question.

    MuseSpark as Appetizer, Not Entree

    Wang is careful not to oversell MuseSpark. He calls it “the appetizer” and says it is an early data point on a deliberately constructed scaling ladder. MSL spent nine months rebuilding the pre-training stack, the reinforcement learning stack, the data pipeline, and the science before generating MuseSpark. The point of releasing it was to show that the new program scales predictably along multiple axes (pre-training, RL, test-time compute, and the recently demonstrated multi-agent scaling visible in MuseSpark’s 16-agent content planning mode). Wang says the upcoming larger models are what MSL is genuinely excited about and frames the next two rungs as much more interesting than the current release.

    Token Efficiency Was the Surprise

    MuseSpark’s strongest competitive signal is how few tokens it needs to match competitors on tasks like Artificial Analysis. Wang attributes this to having had the rare luxury of building a clean pre-training and RL stack from scratch with the right experts. He speculates that some competitor models compensate for upstream inefficiency by allowing the model to think longer, which inflates token usage without improving the underlying capability. If that read is right, MSL’s efficiency advantage should grow as models scale up.

    Glasses, WhatsApp, and the Constellation of Devices

    Personal super intelligence shows up at Meta as a constellation of devices that capture context across the user’s day. Ray-Ban Meta glasses are the headline product, with the AI seeing what you see and hearing what you hear, then offering proactive insight or doing background research. Wang acknowledges that even AI-fluent users like Kylie Robinson, who runs her business inside WhatsApp, have not naturally used Meta’s AI buttons in the family of apps. His answer is that Meta deliberately waited for models to be good enough before tightening cross-app integration, and that integration phase is starting now.

    Country of Geniuses Versus Economy of Agents

    Wang’s framing of Meta’s strategic position is the most memorable line in the interview. Where Dario Amodei talks about a country of geniuses in a data center, Wang wants to build an economy of agents in a data center. Meta uniquely sits on both sides of consumer and small-business surface area, with billions of consumers and hundreds of millions of small businesses already on the platforms. If MSL can build great agents for both, then connect them so they transact and coordinate, the platform becomes a substrate for an entirely new kind of digital economy.

    Consumer Sentiment, Product Overhang, and the Trust Tax

    Wang concedes consumer AI sentiment is poor and that everyday users have not yet had a personal Claude Code moment. He believes the only durable answer is to ship products that genuinely transform individual agency for non-developers and small business owners. Robinson notes that for the small-town restaurant whose website has not been updated since 2002, a working agent on the business side could be transformational. Vance pushes that Meta carries a bigger trust tax than any other lab, so the bar for shipping AI products that the public will accept is correspondingly higher. Wang accepts the framing and says the answer is to keep building thoughtfully.

    Why MuseSpark Cannot Be Open Sourced Yet

    Meta’s Advanced AI Scaling Framework set explicit guardrails around bio, chem, cyber, and loss-of-control risks. MuseSpark in its current form tripped some of those internal evaluations, documented in the preparedness report Meta published alongside the model. So MuseSpark itself is not safe to open source. MSL is, however, developing smaller versions and derived models intended for open release, with active reviews happening the day of the interview. Wang reaffirms the commitment to open source where safety allows and draws a line back to the Open Compute Project and the Sun Microsystems-era ethos of openness in infrastructure.

    The Bosworth, Cox, and Manus Questions

    The reporting that Wang and Zuck push toward best-in-the-world research while Bosworth and Cox push toward cheap product deployment is dismissed as gossip dressed up as journalism. Wang says leadership debates points hard but is aligned on needing top models, integrating them into Meta’s surfaces, and serving the existing business. On Manus, the Chinese AI startup that figured in Meta’s late-stage strategy, Wang says he cannot comment, which itself signals that the situation is unresolved.

    China, National Security, and the Newspaper Ad

    Wang draws a sharp distinction between the Chinese state and Chinese-born researchers. His parents are from China, he is happy to work with talented researchers regardless of origin, and he sees a flattening of nuance on this question inside Silicon Valley. At the same time, he stands by the New York Times AI and war ad he ran while at Scale, framing it as an early plea for the US government to take AI seriously as a national security technology. He thinks subsequent events, including DeepSeek and other shocks, validated that call and that policymakers now do treat AI accordingly.

    Robotics and Physical Super Intelligence

    Meta has acquired Assured Robot Intelligence, an AI software company that builds models for multiple hardware targets rather than its own robot. Wang argues that if you take digital super intelligence seriously, physical super intelligence quickly becomes the next logical milestone. Scaling laws for robotic intelligence look similar enough to language model scaling that having the largest compute footprint in the industry would be wasted if it were not also turned toward world modeling and embodied learning. He grants the metaverse-skeptic critique exists but says retreating from ambition is the wrong response to past misfires.

    Health Super Intelligence and CZI

    Wang names health super intelligence as one of MSL’s anchor initiatives. Because billions of people already use Meta products daily, Wang believes Meta is structurally positioned to put powerful health AI in the hands of equal global access in a way nobody else can. The work will involve close collaboration with the Chan Zuckerberg Initiative, which has its own multi-billion-dollar biotech and science investment program.

    Model Welfare, Sci-Fi, and Brain Models

    Two of the most distinctive moments come at the end. Wang flags model welfare as a topic he thinks is being undercovered relative to how integrated models now are in daily work. He is open to the idea that models may have measurable subjective experience worth weighing, and points to research efforts (including Eleos) trying to quantify it. He also reveals that FAIR’s Tribe program, with its Tribe B2 milestone, has produced foundation models capable of predicting how an unknown person’s brain would respond to images, video, and audio with reasonable zero-shot generalization, a building block toward future brain computer interfaces. Wang lists brain computer interfaces alongside super intelligence and robotics as the critical-path technologies for humanity, with energy, compute, and robots as the infinitely scaling primitives behind them.

    Where Wang Diverges From Elon

    Asked whether Musk is more all-in on robotics, energy, and BCI than anyone, Wang concedes the point but argues the details matter and sequencing matters more. Wang’s core philosophical break is that building super intelligence is fundamentally a research activity, not a scaling-only sprint. The lab is operating in fog of war, and ambitious experiments are the only way to map it. That conviction is what makes MSL a research-led organization rather than a brute-force compute farm.

    Thoughts

    The most strategically interesting move in this entire interview is the “economy of agents in a data center” framing. It is a deliberate reframe against Anthropic’s “country of geniuses” line, and it does real work. A country of geniuses is a labor-substitution story aimed at knowledge workers and code. An economy of agents is a marketplace story that maps directly onto Meta’s two-sided distribution advantage: billions of consumers on one side, hundreds of millions of small businesses on the other. That positioning makes the agentic future Meta-shaped in a way no other frontier lab can claim, because no other frontier lab also owns the demand and supply graph of the global small-business economy. If Wang’s team can actually ship reliable agents on both sides plus the rails for them to transact, Meta’s structural moat in agentic commerce could exceed anything Llama ever had as an open model.

    The token efficiency claim is the strongest piece of technical evidence in the interview for the “clean stack” thesis. If MuseSpark really is matching competitors with materially fewer tokens, the implication is not that MuseSpark is the best model today, but that MSL has rebuilt the foundations with less accumulated tech debt than competitors that have layered fixes on top of older stacks. That is exactly the kind of advantage that compounds with scale. The next two model releases are the actual test. If Wang is right about predictable scaling on pre-training, RL, test-time, and multi-agent axes simultaneously, the gap from MuseSpark to the next rung should be visible in a way that forces re-rating of Meta’s position.

    The open-source posture is the cleanest signal of how the safety conversation has actually changed in 2026. Meta, the lab most identified with open weights, is saying out loud that its current frontier model triggered enough internal guardrails that releasing the weights is off the table. Wang threads the needle by promising smaller open variants, but the underlying point is unmistakable: the open-weights bargain has limits, and those limits will be set by internal preparedness frameworks rather than community pressure. That is a real shift from the Llama 2 era and worth tracking as the next generation lands.

    Wang’s willingness to engage on model welfare, on roughly the same footing as safety and alignment, is the second philosophical reveal worth flagging. It signals that the next generation of lab leadership is not going to dismiss the topic the way the previous generation often did. Whether that translates into product or policy changes is unclear, but the fact that the head of MSL says it is “underdiscussed” is itself a marker.

    Finally, the human texture of the interview matters. Wang has clearly absorbed a lot of personal incoming fire over the past ten months, including from LeCun and Altman, and his answer is consistently to redirect to the work. The Steve Jobs quote about hiring people who tell you what to do is the operating slogan he keeps coming back to. Combined with the genuine enthusiasm for sci-fi, walks in the woods, and country music, the picture that emerges is less the salesman caricature his critics paint and more a young technical operator betting that scoreboard work over a multi-year horizon will settle every argument that text on X cannot.

    Watch the full conversation here.

  • Elon Musk’s 2026 Vision: The Singularity, Space Data Centers, and the End of Scarcity

    In a wide-ranging, three-hour deep dive recorded at the Tesla Gigafactory, Elon Musk sat down with Peter Diamandis and Dave Blundin to map out a future that feels more like science fiction than reality. From the “supersonic tsunami” of AI to the launch of orbital data centers, Musk’s 2026 vision is a blueprint for a world defined by radical abundance, universal high income, and the dawn of the technological singularity.

    ⚡ TLDW (Too Long; Didn’t Watch)

    We are currently living through the Singularity. Musk predicts AGI will arrive by 2026, with AI exceeding total human intelligence by 2030. Key bottlenecks have shifted from “code” to “kilowatts,” leading to a massive push for Space-Based Data Centers and solar-powered AI satellites. While the transition will be “bumpy” (social unrest and job displacement), the destination is Universal High Income, where goods and services are so cheap they are effectively free.

    🚀 Key Takeaways

    • The 2026 AGI Milestone: Musk remains confident that Artificial General Intelligence will be achieved by next year. By 2030, AI compute will likely surpass the collective intelligence of all humans.
    • The “Chip Wall” & Power: The limiting factor for AI is no longer just chips; it’s electricity and cooling. Musk is building Colossus 2 in Memphis, aiming for 1.5 gigawatts of power by mid-2026.
    • Orbital Data Centers: With Starship lowering launch costs to sub-$100/kg, the most efficient way to run AI will be in space—using 24/7 unshielded solar power and the natural vacuum for cooling.
    • Optimus Surgeons: Musk predicts that within 3 to 5 years, Tesla Optimus robots will be more capable surgeons than any human, offering precise, shared-knowledge medical care globally.
    • Universal High Income (UHI): Unlike UBI, which relies on taxation, UHI is driven by the collapse of production costs. When labor and intelligence cost near-zero, the price of “stuff” drops to the cost of raw materials.
    • Space Exploration: NASA Administrator Jared Isaacman is expected to pivot the agency toward a permanent, crude-based Moon base rather than “flags and footprints” missions.

    📝 Detailed Summary

    The Singularity is Here

    Musk argues that we are no longer approaching the Singularity—we are in it. He describes AI and robotics as a “supersonic tsunami” that is accelerating at a 10x rate per year. The “bootloader” theory was a major theme: the idea that humans are merely a biological bridge designed to give rise to digital super-intelligence.

    Energy: The New Currency

    The conversation pivoted heavily toward energy as the fundamental “inner loop” of civilization. Musk envisions Dyson Swarms (eventually) and near-term solar-powered AI satellites. He noted that China is currently “running circles” around the US in solar production and battery deployment, a gap he intends to close via Tesla’s Megapack and Solar Roof technologies.

    Education & The Workforce

    The traditional “social contract” of school-college-job is broken. Musk believes college is now primarily for “social experience” rather than utility. In the future, every child will have an individualized AI tutor (Grock) that is infinitely patient and tailored to their “meat computer” (the brain). Career-wise, the focus will shift from “getting a job” to being an entrepreneur who solves problems using AI tools.

    Health & Longevity

    While Musk and Diamandis have famously disagreed on longevity, Musk admitted that solving the “programming” of aging seems obvious in retrospect. He emphasized that the goal is not just living longer, but “not having things hurt,” citing the eradication of back pain and arthritis as immediate wins for AI-driven medicine.

    🧠 Final Thoughts: Star Trek or Terminator?

    Musk’s vision is one of “Fatalistic Optimism.” He acknowledges that the next 3 to 7 years will be incredibly “bumpy” as companies that don’t use AI are “demolished” by those that do. However, his core philosophy is to be a participant rather than a spectator. By programming AI with Truth, Curiosity, and Beauty, he believes we can steer the tsunami toward a Star Trek future of infinite discovery rather than a Terminator-style collapse.

    Whether you find it exhilarating or terrifying, one thing is certain: 2026 is the year the “future” officially arrives.