Buying a company doesn’t make you that company. That’s the core of Reid Hoffman’s argument against SpaceX’s AI ambitions — and he’s not being subtle. In a recent interview covered by Fortune and Yahoo Finance, the PayPal co-founder and LinkedIn creator called xAI “a complete train wreck“ and labeled SpaceX “a premium-priced CoreWeave” — essentially an expensive GPU landlord dressed up in AI clothing. Worth flagging immediately: Hoffman holds stakes in both OpenAI and Anthropic, making his critique informed but hardly disinterested. That context matters given OpenAI’s sweeping infrastructure ambitions, including the Stargate Project.
The Musk AI Machine Is Burning Through Founders and Credibility
Every original xAI co-founder besides Musk has left, and SpaceX’s post-IPO Cursor acquisition looks more like a patch job than a strategy.
SpaceX went public in June with AI infrastructure as its marquee growth story. Days later, it announced the acquisition of Cursor, an AI coding assistant. Hoffman sees this as Barry Diller’s IAC playbook — the early-2000s internet roll-up strategy where you buy relevance instead of building it. “You’re a premium-priced CoreWeave… which is not an AI company,” he told Yahoo Finance.
The xAI situation is harder to spin. Founded in 2023 with Musk and 11 co-founders, the company has since lost every single one of those original partners:
- Kyle Kosic departed for OpenAI
- Igor Babuschkin pivoted to launching his own venture firm
- By spring 2026, not one of the original eleven remained
Musk has framed some departures as “push not pull,” according to TechCrunch reporting — suggesting firings rather than defections. Grok’s benchmark performance, meanwhile, continues to trail models from Anthropic and OpenAI, adding weight to Hoffman’s “train wreck” characterization.
“As Elon himself has described, it’s a complete train wreck for its kind of building of foundational models.” — Reid Hoffman, according to Fortune.
Two Winners, One Regulatory Mess, and a Generation Caught in the Middle
Hoffman argues OpenAI and Anthropic can both thrive, but regulatory asymmetry and job displacement complicate the picture considerably.
Hoffman rejects the zero-sum framing. Anthropic excels at code generation and is expanding into legal and design applications. OpenAI owns the consumer search layer through ChatGPT, with Codex capabilities Hoffman calls “insufficiently talked about.” He even questions whether Cursor — the tool SpaceX just acquired — has already peaked, now “fading over the horizon” under pressure from Claude Code and Codex. Buying a fading asset doesn’t exactly shore up the AI narrative.
The Anthropic export-control episode troubles him on different grounds. After Amazon CEO Andy Jassy flagged a jailbreak vulnerability in the Fable 5 model, regulators issued an order on June 11, 2026, suspending foreign national access to two Anthropic models. Cybersecurity experts called the response disproportionate, according to Fortune reporting. Hoffman went further, describing the action as “autocratic willy-nilly” — particularly since OpenAI faced no equivalent restriction. For Anthropic, which is preparing for one of the largest anticipated AI IPOs in history, that kind of unpredictable regulatory intervention becomes a new category of investor risk.
For Gen Z, the numbers sting regardless of cause:
- Graduate unemployment has climbed from 3.6% in 2019 to 5.6% in 2026
- 35% of entry-level postings now demand three or more years of experience, per Fortune reporting
- Goldman Sachs’ AI tracker estimated AI was erasing roughly 11,000 net U.S. jobs per month as of mid-2026, with entry-level knowledge roles hit hardest
When AI’s most enthusiastic advocates start drawing sharp lines between real builders and infrastructure cosplayers, you should pay attention. The consolidation isn’t coming. It’s already here.
