SpaceX’s $86 Billion AI Catch‑Up: Buying Cursor and Becoming the Cloud for Its Competitors

Days after its historic public debut, Elon Musk is deploying SpaceX’s newly raised capital on a two‑front AI strategy: buy what you cannot build and rent what you cannot yet beat with. On June 16, 2026, SpaceX locked in a $60 billion all‑stock deal to acquire Anysphere, the company behind the widely used AI coding assistant Cursor . At the same time, the company is selling access to its massive GPU clusters — the Colossus data centers — to Google and Anthropic, the very rivals whose AI models outpace Musk’s own Grok . The plan admits a painful reality inside the company and tries to build an enterprise AI business that does not depend on winning the frontier‑model race outright.

A $60 billion shortcut into coding tools

SpaceX publicly signaled its interest in Cursor in April 2026, when it disclosed an option to either acquire the company for $60 billion or pay $10 billion for their work together . That option converted into a formal acquisition agreement two months later, with the deal expected to close in the third quarter of 2026 . The consideration is entirely in SpaceX Class A common stock, representing about 3.4% dilution based on the valuation from its IPO .

Cursor is one of the most popular AI coding agents, and the acquisition gives SpaceX an immediate foothold in a category that has become one of the first to generate significant enterprise revenue from AI . Musk’s company has been explicit about the rationale: combine Cursor’s product and “reach among expert software engineers” with SpaceX’s Colossus training supercomputer . The resulting enterprise tool is expected to go head‑to‑head with coding assistants from OpenAI, Anthropic, and Google.

Turning Colossus into a revenue engine

While acquiring Cursor is the most visible move, SpaceX’s infrastructure‑leasing business may be even more consequential for its near‑term financials. Before its IPO, the company disclosed plans to lease AI infrastructure to Google for $920 million per month over 32 months — a contract valued at roughly $30 billion in total . Around the same period, Anthropic struck a deal to buy $5 billion worth of GPUs in Colossus 1, and broader reports place Anthropic’s annual spending with SpaceX at about $15 billion across Colossus One and Two .

This compute‑arbitrage approach allows SpaceX to turn its capital‑intensive data‑center buildout into immediate, recurring revenue while its own models catch up. It is a deliberate bet that AI infrastructure demand will remain supply‑constrained, a thesis analysts have described as “building the next‑generation technology foundation” rather than a bubble .

The gap Musk is trying to close

Behind both moves is a stark internal assessment. In early 2026, a SpaceX executive memo sent to xAI employees labeled the training performance “embarrassingly low” and stated plainly that the company was “significantly behind” OpenAI, Anthropic, and others . Musk himself has publicly acknowledged that SpaceX trails the same rivals as well as Chinese open‑source models . Independent ranking frameworks placed Grok well behind ChatGPT, Claude, and Gemini, and one industry observer crystallized the perception as “OpenAI is Coke, Anthropic is Pepsi and Grok is RC Cola” .

Rather than fight only on model quality, SpaceX is building an enterprise AI platform that can compete on distribution, developer tools, and infrastructure. The Cursor deal supplies the application layer that developers interact with daily. The Google and Anthropic contracts fund further capacity expansion and validate Colossus as a neutral compute provider, even for competitors.

An enterprise‑platform bet, not just a model race

Taken together, the two moves shift SpaceX’s AI ambition from being a lagging model lab to becoming an enterprise AI platform. Cursor gives SpaceX a popular end‑user tool for coding, a category where benchmarks improved dramatically over the past year and where business willingness to pay is proven . Renting compute to industry leaders provides decades of cash flow that can subsidize internal model development and further infrastructure growth.

The risk is that SpaceX becomes primarily a picks‑and‑shovels supplier — someone else’s cloud — while the leaders pull further ahead on the AI models that ultimately define the industry. But for a company whose own AI division has admitted it is trailing badly, the war‑chest plan is a deliberate trade: revenue, distribution, and time in exchange for equity and the hope that the gap closes before the music stops.