Weak defenders, computing power rent, IPO support: Why Elon Musk must buy Cursor

On April 21, 2026, when the news broke that SpaceX had reached an agreement with Cursor, Sam Altman was the most silent person in Silicon Valley. This news could have been the biggest tech M&A story of any year, but Altman did not publicly comment.

He had enough reasons to remain silent. Three years ago, he led a $8 million seed round investment in Anysphere, the parent company of Cursor, under the name of OpenAI, which was the company’s first institutional investment. Later, he tried to acquire the company but was rejected. Subsequently, he turned around and acquired Windsurf, a competitor of Cursor, for $3 billion. Then, on April 21, he watched as Musk, at 20 times the price, locked in the asset he once desired most.

Six days later, he was set to meet Musk in a courtroom in Oakland, California.

This $60 billion deal is almost universally read at face value. But behind it are five layers of systematically obscured truths: Musk is not a strongman, $10 billion is not a breakup fee, integration predated the announcement, compute rent predated the agreement, and the entire game serves a larger IPO bill.

Pricey Defense

Starting with the layer “Musk is not a strongman.”

xAI currently owns the world’s largest AI computing cluster, the Colossus supercomputing center, with 100k Nvidia H100 GPUs, planning to expand to 200k. This is the number most frequently cited in mainstream reports about the company, and it’s the figure that most easily fuels the intuition that “Musk has already won the compute war.”

But shortly after Colossus was built, an internal memo quietly surfaced to the media. The memo was written by Michael Nicolls, who had just been appointed president of xAI after leaving his position as Senior Vice President of Starlink Engineering at SpaceX. In the memo, he used a stern tone to judge that xAI was “significantly lagging” in AI competition, supported by specific figures: xAI’s model floating-point utilization (MFU) was about 11%, while the industry average was 35% to 45%.

The Colossus supercomputing center in Memphis, Tennessee

This means that over 60% of the 100k H100 GPUs are idle. The world’s largest compute cluster is actually operating at less than one-third of industry average efficiency.

The timing of this memo’s delivery is even more embarrassing given the context. All 11 co-founders of xAI have left, Musk publicly admitted that xAI “was not built well the first time and is being rebuilt from scratch,” and the company’s most ambitious all-in-one AI agent project, “Macrohard,” has stalled due to the departure of its core leader. Nicolls was brought in precisely to transplant SpaceX’s extreme efficiency engineering culture, to get that massive machine running at high speed again.

Locking in Cursor for $60 billion is not a casual acquisition by a strong player expanding territory. It’s a company in the process of rebuilding technologically, with severely low compute efficiency, using money to buy time, while also seeking a sufficiently large commercial load for that idling machine. All of this was already underway 40 days before the official announcement.

Ginsberg and his partner Andrew Milich are co-heads of Cursor’s product engineering, jointly responsible for the architecture and iteration of all core product features. During Cursor’s rapid growth from zero to $2 billion in annual revenue, setting SaaS records, they were the two most critical figures on the product side. On March 12, they both announced they were joining xAI, reporting directly to Musk, with the task of rebuilding the programming capabilities of the Grok model from scratch.

Andrew Milich left a remark when announcing his joining xAI: “Almost 10 years ago, I interned at SpaceX, working on the cockpit display system for Dragon 2.” Ginsberg’s return’s significance can only be understood in the context of his full trajectory.

The cockpit display system of Dragon 2 was one of SpaceX’s most core human-machine interaction projects at the time, with only a handful of interns involved. After leaving SpaceX, he and Milich co-founded Skiff, an end-to-end encrypted document collaboration platform, which was acquired by Notion in February 2024. Later, the two joined Cursor, pushing it to the top of the AI programming market. Now, armed with years of product judgment and technical architecture experience accumulated at Cursor, he is returning to the starting point of this path.

When you expand this timeline: Ginsberg’s return, Milich’s joining, and both being explicitly tasked with “rebuilding Grok’s programming capabilities,” the core part of the acquisition was already completed 40 days before the official announcement. The contract is the endpoint, not the starting point.

The True Bill of $10 Billion and the Precise Courtroom Targeting

The structure of the acquisition deal is designed as follows: at some point later this year, SpaceX must make a binary decision—either exercise the $60 billion acquisition option or pay $10 billion as a “partnership fee.” If the acquisition is ultimately not completed, the $10 billion is a one-time fee to settle the partnership.

Media reports generally refer to this $10 billion as a “breakup fee,” but this definition fundamentally obscures its true nature. To understand why Cursor accepted such a peculiar structure, one must first understand the predicament it faces.

Cursor’s origins relied on the underlying models provided by Anthropic. Claude is its core technology foundation. Then Anthropic launched Claude Code, a coding AI tool directly competing for the same user base as Cursor, turning the supplier into a rival. But that’s not the worst part.

Internal analysis at Cursor shows that Anthropic’s Claude Code subscription costs up to $200 per month, but the actual compute cost per heavy user is as high as about $5,000. Last year, this figure was $2,000, and it increased by 1.5 times within a year. This means Anthropic is delivering this service at a loss of about $4,800 per user per month, subsidizing the service with venture capital.

They are using venture capital ammunition to wage a pricing war that leaves competitors economically stranded. Any company relying on Anthropic’s models and offering similar experiences cannot cover API costs with subscription fees.

Faced with this situation, Cursor’s CEO Michael Truell said: “Our strategy is to use the best technology from partners and our own developed technology in a comprehensive way.” This is the clearest statement a founder can make publicly. On the other hand, Cursor must train its own models, which requires finding a compute source that does not depend on competitors.

xAI’s Colossus cluster is currently the largest independent compute resource available to Cursor. The $10 billion “partnership fee” encapsulates the prepaid value of this compute leasing relationship. Cursor’s predicament has also become a fitting piece in a larger game.

The relationship between OpenAI and Cursor spans a full three-year arc, but mainstream reports have never fully pieced it together.

The story begins in October 2023. OpenAI led an $8 million seed round investment in Anysphere, becoming Cursor’s first major institutional investor. At that time, Cursor was a nascent AI programming tool company, and OpenAI’s investment served both as funding and as a form of endorsement.

Two years later, the situation had completely reversed. In November 2025, Cursor completed a $2.3 billion Series D funding round, with a valuation of $29.3 billion. OpenAI began seriously discussing the possibility of acquiring Cursor, wanting to buy back the seed investment. Cursor refused. Subsequently, OpenAI turned to acquire Windsurf, a competitor of Cursor, for $3 billion, formerly known as Codeium. The unicorn valued at $100k was rejected, and OpenAI could only settle for a less optimal choice at one-tenth the price.

Then, on April 21, SpaceX announced the $60 billion deal, just six days before Altman and Musk’s court confrontation in Oakland. The lawsuit stemmed from Musk accusing Altman of betraying OpenAI’s original nonprofit mission, making it the most publicized confrontation of longstanding grievances.

There’s no evidence that April 21 was deliberately chosen, but the dramatic timing cannot be ignored: six days before the court battle, Musk publicly locked in the company that OpenAI had once invested in and later refused to acquire, at $60 billion. The seed investment made three years ago with $8 million is now worth $60 billion. Altman is well aware of this calculation.

And Musk’s calculation is not just aimed at Altman.

The Groundwork for IPO

On April 1, 2026, SpaceX secretly filed for an IPO with the U.S. Securities and Exchange Commission, targeting a valuation of $1.75 trillion. The core narrative supporting this figure is SpaceX’s grand plan to deploy up to 1 million data center satellites in space, replacing terrestrial data centers’ power and water cooling needs with solar energy and natural space heat dissipation, providing cheaper infrastructure for AI computation. Musk has claimed on multiple occasions, “Space will be the cheapest place to run AI.”

But according to the S-1 documents leaked by Reuters and other media, SpaceX admits in its legal filings that this plan involves “unproven technology,” and its commercial viability is uncertain. The company’s official assessment of its biggest AI gamble is a question mark.

This question mark explains the real logic behind acquiring Cursor. Whether SpaceX’s space compute plan can succeed depends on unverified physical and engineering assumptions. Cursor generates 150 million lines of code daily for global enterprises—an actual, ongoing, large-scale demand. If the space compute plan’s future remains uncertain, then the most critical task is to find a sufficiently large and stable commercial load for the idle Colossus cluster on the ground.

With $2 billion in annual revenue, Cursor is one of the fastest-growing SaaS products worldwide. Integrating it into SpaceX’s ecosystem not only provides an outlet for compute monetization but also offers a real-world example to the capital markets that “AI infrastructure has landed.” This is more convincing in IPO roadshows than any space data center outlook.

Acquiring Cursor is SpaceX’s way of using a certain software asset to underpin an uncertain hardware vision.