SpaceX Q1 Earnings: $15.8B AI Spend, Stock Drops 8% on Lockup Fears
SpaceX burned $15.8B on AI in its first public quarter. Stock fell 8% as lockup expires and Musk promises Moon robots. What operators need to know now.
What Happened
SpaceX delivered its first earnings report as a publicly traded company on Tuesday, August 4, 2026, and the numbers were staggering. According to Gizmodo, the company spent $18.4 billion in capital expenditure during its first quarter — with $15.8 billion of that directed entirely toward AI infrastructure. The AI division ended the quarter with a $1.3 billion net operating loss.
The market's response was swift: SpaceX stock dropped more than 8% in after-hours trading. The sell-off was compounded by the imminent expiration of the company's first lockup period on Thursday, August 7, which will allow insiders to sell more than $100 billion worth of stock for the first time since the IPO.
Musk used the earnings call to justify the spending with a series of ambitious promises. He claimed SpaceX would end 2026 with over two gigawatts of compute capacity, built exclusively on Nvidia chips. He announced plans to begin launching Starmind AI satellites in 2027 as part of a broader vision to place up to one million AI data centers in Earth's orbit. Grok 5, he said, would be unveiled before year-end and trained on "all the data that SpaceX has ever produced," making it "by far the best engineer." Musk also claimed Grok 5 would reach AGI-level capabilities.
Beyond AI, Musk promised a rocket launch cadence of once per day by next year and robots working in manufacturing on the Moon.
Why It Matters
This earnings report provides the first hard financial window into SpaceX's AI strategy since the company went public in mid-June. The $15.8B AI spend in a single quarter is extraordinary — it exceeds the annual AI infrastructure budgets of most major technology companies. The $1.3B operating loss confirms that SpaceX is subsidizing its AI push with no near-term profitability in sight.
For the broader AI market, this matters on multiple fronts. First, SpaceX already disrupted API pricing in July with Grok 4.5, launched at half the price of competitors like Anthropic and OpenAI. That pricing now makes more sense: SpaceX is willing to absorb massive losses to gain market share. But the stock market's 8% drop signals that investors have limits on how much red ink they'll tolerate, especially with $100B+ in insider shares about to unlock.
If the lockup expiration triggers significant selling and the stock continues to decline — it's already well below its IPO price — SpaceX's cost of capital for continued AI investment rises. That could slow the compute buildout Musk promised, which would affect Grok API availability, pricing, and the competitive dynamics of the entire AI inference market.
Who Is Affected
AI startup founders relying on Grok APIs for cost-effective inference should assess their dependency. SpaceX's willingness to lose billions quarterly means pricing could stay low, but a capital squeeze from a falling stock price could change that calculus rapidly.
Developers and operators building on AI APIs now have another data point on the financial sustainability of their vendors. SpaceX's $1.3B quarterly AI loss is a reminder that aggressive pricing often masks unsustainable economics.
Public market investors and enterprise AI buyers evaluating vendor stability should note that SpaceX's stock is already below its IPO price, with further pressure likely from lockup selling. Vendor financial health is a real operational risk.
Strategic Implications
For AI Startup Founders
SpaceX's massive AI spend and willingness to operate at a loss means Grok API pricing may remain artificially low for the foreseeable future. However, founders should avoid building critical infrastructure on a single vendor whose parent company's stock is declining and whose capital position could tighten. Diversify your API providers and treat Grok's pricing advantage as a bonus, not a foundation.
For Developers Building with AI APIs
Grok 5 is promised before year-end with claims of AGI-level engineering capabilities. Treat this as a directional signal, not a commitment — Musk's historical delivery on timelines is optimistic at best. Plan around Grok 4.5 as your current production ceiling and benchmark alternatives. The Starmind satellite and orbital data center plans are fascinating but years away from operational relevance.
For Non-Technical Business Owners
The stock market's negative reaction to SpaceX's AI spending is a useful signal: even with Musk's track record, markets are skeptical of AI capex without clear monetization. If you're choosing an AI vendor, factor in financial stability alongside pricing and capabilities. A vendor losing $1.3B per quarter on AI may not be a reliable long-term partner if capital markets turn hostile.
What to Watch Next
Monitor Thursday's lockup expiration and the volume of insider selling — that will be the strongest near-term signal of whether SpaceX's stock stabilizes or enters a deeper decline. Also watch for the official earnings release (beyond Gizmodo's reporting) to verify the financial figures, and for any competitive pricing response from OpenAI or Anthropic in the wake of SpaceX's disclosed AI losses.
Frequently Asked Questions
Q: How much did SpaceX spend on AI in its first quarter as a public company?
A: According to Gizmodo's reporting on SpaceX's first earnings call, the company spent $15.8 billion of its $18.4 billion total Q1 capex on AI infrastructure, resulting in a $1.3 billion net operating loss for the AI division.
Q: When does SpaceX's lockup period end and what does it mean?
A: SpaceX's first lockup period ends on Thursday, August 7, 2026. This allows company insiders to sell more than $100 billion worth of stock for the first time since the IPO, which could increase selling pressure on an already-declining stock price.
Q: What did Musk promise about Grok 5?
A: Musk said Grok 5 would be unveiled before the end of 2026, trained on all data SpaceX has ever produced, and would be "by far the best engineer." He also claimed it would reach AGI-level capabilities, though this remains an unverified claim.