Reuters: SpaceX IPO shows how expensive the AI race has become

Reuters reports that SpaceX’s new IPO filing makes one thing clear: AI is no longer a side project inside Elon Musk’s space company. It has become a capital-heavy core bet with large losses, large infrastructure plans and unusually concentrated control.

According to Reuters, SpaceX is targeting a valuation of $1.75 trillion. That would make the listing one of the largest in U.S. history. But beneath that headline sits the more relevant AI story for boards, CIOs and CFOs: xAI and data-center capacity are moving AI competition away from model choice and into capital, power, contracts and governance risk.

The numbers are unusually concrete. Reuters says SpaceX generated $4.69 billion in first-quarter revenue and booked a $1.94 billion operating loss. The AI division alone accounted for $2.47 billion of losses on $818 million of revenue. The acquisition of xAI also accounted for 76% of SpaceX’s $10.1 billion in capital spending in the quarter.

This is not just a Musk story. It points to the direction of the broader AI market. The strongest AI companies are not only building models. They are building, financing or locking up data centers, power agreements, chip capacity, networks, security zones and long-term customers. Once AI services are sold with guaranteed capacity, low latency and dedicated agent environments, infrastructure becomes part of the product.

AI is becoming a capital question

Reuters says SpaceX’s outlook relies partly on markets that are not yet mature, including AI data centers in space. The filing refers to a potential $28.5 trillion market. That should be read as ambition, not forecast. But the direction matters: SpaceX is not trying to be valued only as a rocket or satellite company. It is trying to be valued as an AI infrastructure platform.

For CIOs and boards, the key question is not whether SpaceX will build space-based data centers. The point is that AI vendors are increasingly competing on control of capacity. Reuters also reports that SpaceX, through its AI infrastructure platform, has deals under which Anthropic will pay $1.25 billion a month to use compute capacity from the Colossus and Colossus II clusters in Memphis through May 2029.

That sets a new benchmark for the obligations that can sit behind enterprise AI. When a model or agent platform is offered to companies, it may be backed by multi-year compute contracts, specific data-center locations, dedicated energy arrangements and financing needs that must be recovered through customer contracts. That affects pricing, lock-in and supplier risk.

Power is no longer background noise

Wired reported the same evening that SpaceX has committed more than $2.8 billion to buying gas turbines for AI data centers. According to Wired, the spending comes as the use of those turbines has triggered complaints, lawsuits and regulatory questions over emissions and permits.

This is not an environmental footnote. Power access has become a hard constraint on AI buildout. When vendors need temporary or dedicated power generation to keep pace, the risk of delays, political resistance, cost increases and reputational damage rises. For enterprises buying AI capacity, this belongs in supplier due diligence. Where is the capacity located? What energy and regulatory risks are attached? What happens to prices if the vendor must finance new power and new data centers faster than planned?

Governance matters as much as model quality

Reuters also points to governance. Musk is set to retain 85.1% of voting control, and the company uses a share structure that gives ordinary investors limited influence. At the same time, Reuters says SpaceX disclosed that it has been named as a defendant in multiple lawsuits tied to Grok’s image-generation and editing features.

For enterprise buyers, this is not mainly an IPO story. It is a reminder that AI platforms must be assessed as strategic suppliers, not just software. Model quality, API price and demos are only part of the picture. Capital needs, control structures, contract terms, data handling, geographic exposure, legal disputes, power access and exit options matter just as much.

Norwegian and European leaders should read the SpaceX filing as a warning about the next phase of the AI market. Compute is becoming more expensive, more political and more concentrated. Vendors that can promise capacity may gain stronger negotiating power. AI procurement therefore has to be treated as infrastructure and financial risk, not as an ordinary SaaS subscription.

The practical checklist is short: require clarity on where critical AI capacity runs, how long the vendor is tied to underlying compute contracts, how pricing can change during capacity shortages, what data can be moved out, and how the company can switch model or platform without losing processes, logs and security controls.

The SpaceX story is about Musk, but the implication is larger than one company. The AI industry is building the factory floor while customers are already moving workflows inside. That can be powerful when everything works. It becomes expensive when capacity, power or governance breaks.

Sources and media

Primary source: Reuters, "SpaceX IPO filing lays bare losses and Musk control as it stakes future on AI", published May 20, 2026: https://www.reuters.com/legal/transactional/bound-mars-elon-musks-spacex-unveils-filing-blockbuster-ipo-2026-05-20/

Additional source: Wired, "SpaceX Is Spending $2.8 Billion to Buy Gas Turbines for Its AI Data Centers", published May 20, 2026: https://www.wired.com/story/elon-musk-spacex-spending-gas-turbines-grok/

Thumbnail: OpenAI Image 2 / hogby.ai. Editorial illustration generated by hogby.ai, not an actual SpaceX, xAI or Reuters graphic.