SpaceX's Coding Ambitions Hit a Wall as Cognition CEO Flatly Rejects Acquisition Talk
Key takeaways
- Cognition CEO Scott Wu denied Bloomberg's report that SpaceX sought to acquire the AI coding startup.
- SpaceX recently closed a $60 billion acquisition of Cursor and jointly launched the Grok 4.6 coding model.
- Cognition raised $1 billion at a $25 billion valuation in May and is reportedly now targeting a $40 billion raise.
Cognition CEO Scott Wu took to X on Wednesday to dispute a Bloomberg report claiming SpaceX had approached the AI coding startup about an acquisition, writing that the story was inaccurate and that Cognition 'is not for sale.' Bloomberg cited sources familiar with the matter when it published the report, though Wu said the two companies had not been in any talks. The denial was swift and unambiguous, though notably Wu did not address a secondary Bloomberg claim that the two companies are still discussing a potential compute arrangement. SpaceX has not responded to requests for comment.
The Bloomberg story arrived just days after SpaceX completed its $60 billion acquisition of Cursor, another AI coding platform, with Cursor and SpaceX jointly releasing Grok 4.6 — a model optimized for coding and multi-step agentic tasks — earlier this month. SpaceX absorbed Elon Musk's AI company xAI earlier this year, then rode a blockbuster IPO in June to a peak market capitalization near $2.3 trillion. Despite that financial momentum, the xAI operation is widely regarded as behind rivals like Anthropic, Google DeepMind, and OpenAI in model capability and enterprise adoption.
Musk has been vocal about AI's future role inside SpaceX, telling employees last week that AI will represent roughly 99% of the company's value within four or five years. Reaching that goal demands serious revenue from AI products, and coding tools have emerged as one of the clearest commercial paths — Anthropic's rapid growth, driven substantially by Claude Code, being the most prominent example. Acquiring Cognition and its flagship coding agent Devin, which counts Mercedes-Benz, Citi, and Goldman Sachs among its enterprise clients, would have added meaningful scale to SpaceX's coding ambitions.
Cognition itself has been through a turbulent stretch. The startup acquired the remaining assets of Windsurf after Google DeepMind struck a $2.4 billion deal to acqui-hire that company's CEO and top researchers. Following the merger, Cognition laid off 30 employees and offered buyouts to Windsurf's remaining 200 staff, with those who stayed reportedly expected to work more than 80 hours a week, six days in the office. Despite the internal turbulence, Cognition raised a $1 billion funding round in late May at a $25 billion post-money valuation, and Bloomberg now reports the company is in early talks for a new round at a $40 billion valuation.
Bloomberg noted that acquisition discussions, if they occurred at all, are no longer active, though a potential compute partnership — where Cognition would use SpaceX's data center capacity — may still be on the table. SpaceX has been selling compute access to other AI players, including Anthropic, as it builds toward eventually housing data centers in orbit. Whether Cognition becomes a customer, a partner, or remains fully independent is still an open question, but Wu's public statement makes one thing clear: the company is not entertaining a buyout.
The bigger picture
SpaceX's acquisition of Cursor for $60 billion already represented a dramatic escalation in the AI coding wars, but the Bloomberg report — and Wu's quick rebuttal — reveals just how aggressively SpaceX is scanning the landscape for further consolidation targets. Cognition's denial may be genuine, or it may be a negotiating posture; the fact that Wu left the compute-arrangement angle unanswered is worth noting. Either way, the episode spotlights a broader pattern: AI coding startups have become among the most coveted assets in tech right now, and every major AI model maker is trying to lock in developer tooling before the market matures.
For rivals like Anthropic and Google DeepMind, SpaceX's hunger for coding assets creates real competitive pressure. Anthropic already has Claude Code gaining enterprise traction, and Google DeepMind secured Windsurf's research leadership for $2.4 billion. If SpaceX can successfully integrate Cursor and eventually pair it with expanded compute infrastructure — possibly including orbital data centers — it could carve out a distinctive enterprise pitch that pure model companies can't easily replicate. Grok's ongoing controversies, including the MechaHitler episode and nonconsensual imagery scandals, remain a reputational drag on enterprise sales, which makes owning best-in-class coding tools more strategically important, not less.
Cognition's reported $40 billion valuation target is the number readers should watch closely. If that round closes at that figure, it would signal that independent AI coding companies still have room to grow without a corporate parent — and that acquirers will need to pay a steep premium to change that calculus. For SpaceX, the smarter near-term play may be the rumored compute deal, which would create dependency and data leverage without requiring a full buyout. The next few months of funding announcements and partnership disclosures will tell us a great deal about where AI coding consolidation is actually heading.
We're covering this one closely because the SpaceX-Cognition story sits at the intersection of several threads we've been tracking all year: the race to monetize AI through developer tools, Elon Musk's increasingly sprawling AI empire, and the shrinking pool of independent coding startups. When a founder publicly denies an acquisition report within hours of publication, that's almost always newsworthy on its own — it tells you something about how sensitive these companies are right now to being seen as available or vulnerable. The $40 billion valuation figure that Bloomberg dropped into the same story suggests Cognition is very deliberately positioning itself as expensive and self-sufficient, which is its own kind of signal. We think readers building an understanding of the AI infrastructure landscape — who owns what, who's selling compute to whom — will find this moment genuinely instructive. Watch the compute partnership angle especially; that's the thread that could develop into something much bigger.
As an Amazon Associate, LagPing earns from qualifying purchases. Product links are affiliate links.
You might also like
Decoding AI's Growing Vocabulary: What OpenAI's New Reasoning Models Mean for the Field
Sep 8
Kalanick's Atoms Signals AV Ambitions With Hiring Blitz and Uber Talks
Sep 7
ZeWood's Laser-Cut Middle-Earth Wall Art Is the Tolkien Gift Nobody Knew They Needed
Aug 5
Superblocks Locks Vibe Coding Inside Enterprise AWS Clouds — And It's a Sign of What's Coming
Aug 4