Reid Hoffman has spent years watching Silicon Valley from the within — as a founder, an investor, and a decade-long Microsoft board member. So when he says SpaceX’s AI technique quantities to purchasing relevance quite than constructing functionality, and describes xAI as “a whole practice wreck,” it lands with the load of somebody who truly is aware of the place the our bodies are buried.
Key takeaways
- Reid Hoffman says SpaceX isn’t an AI firm and criticizes its acquisition technique as an try to “purchase your method into relevance.”
- All 11 of xAI’s authentic co-founders had left the corporate by Could 2026, signaling deep organizational instability.
- SpaceX went public on June 12, 2026 with AI on the middle of its IPO story, then rapidly acquired Cursor — an AI coding instrument Hoffman believes has already peaked.
- Reflection AI signed a deal to pay SpaceX $150 million per 30 days beginning July 1, 2026, for entry to Nvidia GB300 chips on the Colossus 2 knowledge middle — a deal value as much as $6.3 billion by means of 2029.
- The U.S. authorities issued an export management order on June 11, 2026, suspending international entry to Anthropic’s Fable and Mythos fashions — a transfer Hoffman known as “autocratic willy-nilly.”
Reid Hoffman on SpaceX and xAI: shopping for the ticket, lacking the trip
Talking on Rana el Kaliouby’s Pioneers of AI podcast, Hoffman was direct: “SpaceX isn’t an AI firm.” His argument isn’t about rockets or Starlink — it’s about what it takes to truly compete on the frontier of synthetic intelligence. Leasing compute and snapping up AI startups, in his view, isn’t the identical factor as constructing foundational fashions that transfer the sphere.
His comparability was pointed. “You possibly can nearly consider it because the IAC of AI,” he mentioned, invoking Barry Diller’s internet-era acquisition machine — an organization that used market capitalization to roll up digital properties with out ever proudly owning the underlying know-how. “Use the market cap to purchase AI firms and attempt to purchase your method into relevance.” And for the compute leasing enterprise particularly, Hoffman put it extra bluntly: “You’re a premium-priced CoreWeave. I get it. Which isn’t an AI firm.”
The critique lands more durable whenever you consider what SpaceX has truly been doing. The corporate went public on June 12, 2026, with AI woven by means of its IPO narrative. Inside days, it introduced the acquisition of Cursor, the AI coding instrument. And its Colossus 2 knowledge middle — initially constructed by xAI for its personal mannequin coaching — is now functioning as a business compute platform, with Anthropic paying $1.25 billion per 30 days, Google paying $920 million per 30 days, and now Reflection AI committing $150 million per 30 days beginning July 1, 2026 by means of 2029. That deal alone is value as much as $6.3 billion if it runs to time period, with both social gathering in a position to exit with 90 days’ discover after the primary three months.
That’s actual income. But it surely’s infrastructure income — the type that makes you a landlord within the AI race, not a participant in it.
The xAI collapse Hoffman didn’t sugarcoat
xAI’s troubles transcend strategic positioning. Hoffman known as it “a whole practice wreck for its sort of constructing of foundational fashions,” noting the corporate was on its “third restart.” The numbers behind that evaluation are stark: by Could 2026, all 11 of xAI’s authentic co-founders had departed the corporate. The wave started in earnest in February when Tony Wu, thought of some of the operationally central figures, introduced his resignation. Restructuring adopted, however the departures saved coming.
The organizational exodus issues past the human curiosity angle. Co-founder continuity is among the strongest early indicators of analysis tradition and mannequin high quality. When each authentic architect has left, the institutional information and technical path they embodied leaves with them. Grok, xAI’s flagship mannequin household, has confronted persistent criticism for lagging behind Anthropic and OpenAI on benchmark efficiency — and the co-founder turnover makes a significant course correction more durable, not simpler.
Reflection AI’s Colossus deal and what it reveals about SpaceX
The Reflection AI compute settlement — confirmed by each TechCrunch and CNBC — provides one other dimension to the SpaceX AI infrastructure story. Reflection, based in 2024 by two former Google DeepMind researchers and final valued at $25 billion, is constructing open-weight AI fashions and dealing with authorities shoppers together with the Division of Power’s Genesis Mission and broader Pentagon AI packages.
Its pitch is intentionally timed. After the U.S. authorities suspended international entry to Anthropic’s closed Fable and Mythos fashions, open-source AI gained a brand new and concrete argument: if a authorities can shut down a closed mannequin in a single day, perhaps enterprises and nations ought to suppose more durable about dependence on proprietary programs. Reflection has leaned immediately into that narrative, positioning itself as “American open intelligence.”
For SpaceX, every new compute deal reinforces the identical sample Hoffman is critiquing — AI infrastructure as income, not AI functionality as a aggressive benefit. The corporate now has Anthropic, Google, Cursor, and Reflection as compute prospects. That’s a formidable business place. Whether or not it constitutes an AI technique is a distinct query solely.
Anthropic, export controls, and what unpredictable regulation truly prices
On June 11, 2026, the U.S. authorities issued an export management order suspending all international nationwide entry to Anthropic’s Fable and Mythos fashions. The set off, in response to reporting from Fortune, was Amazon CEO Andy Jassy elevating considerations a few found jailbreak within the Fable 5 mannequin — a vulnerability Anthropic itself had been working to repair. Cybersecurity specialists extensively known as the federal government’s response disproportionate.
Hoffman, who’s an investor in Anthropic, landed in an identical place — however his concern went past the particular resolution. “It doesn’t appear like there’s something that’s a selected principled, right here’s-the-way-that-we’re-navigating-through-things, apply-kind-of-a-rule-of-law-and-predictability,” he mentioned. The sensible translation: the federal government gave the impression to be “hitting them with a stick” due to pre-existing friction with the corporate, not due to a transparent and persistently utilized framework.
He known as the strategy “autocratic willy-nilly” and “very sub-optimum.” The asymmetry — Anthropic penalized whereas OpenAI was not — is what troubled him most. And the implications attain past Anthropic’s speedy enterprise. For an organization making ready what is anticipated to be one of many largest IPOs in historical past, regulatory motion of this sort creates a brand new and poorly-priced danger class for buyers: not market danger, not technical danger, however the danger {that a} authorities intervention can droop your product line with minimal discover and opaque reasoning.
That danger, now documented, will likely be priced into future AI IPOs in methods it wasn’t earlier than.
Why Hoffman sees room for each Anthropic and OpenAI — and the place Cursor suits
Hoffman pushed again firmly on the zero-sum framing that dominates AI protection. As an investor in each Anthropic and OpenAI, he argued the 2 firms occupy meaningfully totally different aggressive lanes. Anthropic, in his view, is strongest in code and increasing into design and authorized purposes. OpenAI and ChatGPT perform extra like a consumer-facing search interface, whereas its Codex coding product is “insufficiently talked about” given its precise capabilities.
The pointed query he raised about Cursor — the AI coding instrument SpaceX simply acquired — was telling. “Cursor appears to have had its vibrant star some variety of months in the past and appears to be fading over the horizon,” he mentioned. Cursor has confronted mounting aggressive stress since early 2026 as Claude Code and Codex gained floor, with builders more and more questioning whether or not a standalone AI coding IDE nonetheless instructions a premium. If Hoffman is true, SpaceX acquired a instrument at or previous its peak — one other knowledge level for the “shopping for relevance” critique.
Gen Z, the job market, and Hoffman’s counter-narrative
On the query of AI and employment, Hoffman’s recommendation to youthful staff reduce in opposition to the prevailing nervousness. “I’ve been interested by writing an essay on the sort of errors which can be made by faculty graduates booing or in any other case dissing AI,” he mentioned. His prescription: deal with the second as an entry benefit, not a menace. “You guys have the chance to be technology AI — the place you come into the workforce saying, ‘I do know this quite a bit higher than all of you.’”
The information he’s pushing in opposition to is actual. A Goldman Sachs AI tracker discovered that by April 2026, AI was erasing roughly 16,000 web U.S. jobs per 30 days, with Gen Z bearing a disproportionate share as entry-level information roles face the best displacement danger. Graduate unemployment rose from 3.6% in 2019 to five.6% in 2026. By mid-2026, 35% of entry-level job postings required no less than three years of expertise, and 45% of firms have been utilizing automated rejection programs at early hiring phases.
Hoffman’s counter is that almost all of this ache is being misattributed to AI when the precise drivers are pandemic over-hiring, distant work normalization, and world financial turbulence. His e book Superagency frames the prescription as an company mindset: “The AI is my instrument, companion, automotive, et cetera, as I navigate issues. The AI can do a complete bunch of fantastic issues itself however isn’t full — and people can add in loads of vital and necessary issues.”
Manas AI and what comes subsequent for Hoffman
Hoffman’s personal subsequent chapter is furthest from the infrastructure and compute debates dominating the present AI dialog. After stepping again from the Microsoft board — a tenure that included facilitating the LinkedIn acquisition, the GitHub buy, and the early Microsoft-OpenAI partnership — he’s now targeted on Manas AI, his drug discovery firm co-founded with Ujjwal and Sid.
Manas is producing small molecule proposals that the corporate’s computational chemists describe as “genuinely promising” — Hoffman says that evaluation was the set off for his resolution to go all-in. The corporate’s pitch deck, he revealed, describes it as “an AI drug discovery manufacturing facility for creating monopolies” — legally permissible as a result of pharmaceutical IP is a government-sanctioned monopoly by design.
It’s the longest-horizon guess in a profession outlined by them. And notably, it’s the one play in his portfolio that has nothing to do with whether or not SpaceX is or isn’t an AI firm — a debate that, given the Colossus compute offers now accumulating, the market might resolve quicker than Hoffman’s critique suggests.
FAQ
Why does Reid Hoffman say SpaceX isn’t an AI firm?
Hoffman’s argument is that SpaceX’s AI involvement is primarily in leasing compute infrastructure and buying AI startups, not in constructing core AI fashions or advancing frontier analysis. He compares the strategy to an acquisition roll-up technique — shopping for market presence quite than incomes it by means of technical functionality.
What points is xAI going through in response to Reid Hoffman?
Hoffman describes xAI as “a whole practice wreck” when it comes to constructing foundational fashions, and notes the corporate has gone by means of a number of restarts. Probably the most concrete indicator of instability: by Could 2026, all 11 of xAI’s authentic co-founders had left the corporate.
What’s the U.S. authorities’s stance on Anthropic’s AI fashions?
On June 11, 2026, the U.S. authorities issued an export management order suspending international nationwide entry to Anthropic’s Fable and Mythos fashions, citing safety considerations together with a found jailbreak vulnerability within the Fable 5 mannequin. Hoffman criticized the motion as missing principled justification and known as the regulatory strategy “autocratic willy-nilly.”
What recommendation does Reid Hoffman give to Gen Z about AI and careers?
Hoffman urges Gen Z to embrace AI as knowledgeable benefit quite than a menace. His view is that younger staff who develop deep AI fluency can place themselves as important guides for organizations making an attempt to change into AI-native — turning a second of disruption right into a profession entry level.
Article produced with the help of synthetic intelligence and reviewed by the editorial staff.