As technology companies prepare to report results this week, investors are reassessing whether artificial intelligence spending can generate returns commensurate with the capital being deployed. SpaceX's data center expansion, financing plans and compute pricing will offer important clues for valuing the business, while CoreWeave's recent share-price decline has heightened sensitivity to the debt and funding needs of AI infrastructure companies.
SpaceX targets 10 GW by end-2027
SpaceX is expected to hold an earnings call with Wall Street analysts after Tuesday's close, with its AI data center build-out likely to be a central topic. Third-quarter AI revenue is projected to rise 230% year on year to $3.78 billion, although it still represents a relatively small share of SpaceX's total revenue. The business currently earns most of that revenue by renting data center capacity to customers, rather than licensing AI models or selling software subscriptions.
Elon Musk has said SpaceX plans to increase AI data center capacity from 1.4 gigawatts in mid-2026 to 10 gigawatts by the end of 2027, roughly seven times the current level. In a post on social media on Monday, Musk said the eventual scale could be “more than 10 gigawatts.” Data centers are commonly measured by their power capacity, so a gigawatt-scale expansion implies sustained increases in infrastructure investment and electricity demand.
SpaceX is seeking financing of up to $40 billion led by Apollo Global Management. Edison Yu, an analyst at Deutsche Bank, has identified compute deployment as one of the biggest variables affecting SpaceX's earnings outlook for next year. The company's main supercomputing clusters are concentrated around Memphis, Tennessee, while it is also building data centers in Texas. Debate over local power infrastructure is prompting some states to reassess data center expansion plans.
GPU pricing and customer contracts in focus
SpaceX has become a major Nvidia customer. Investors will be watching for details on the prices SpaceX charges for GPU rentals and the terms governing contracted compute capacity. Those disclosures could also influence assessments of profitability across the wider AI infrastructure sector.
Anthropic is considered one of SpaceX's largest data center customers. The agreement is reportedly worth as much as $84.5 billion through 2029. Google is a newer SpaceX customer. SpaceX has disclosed that Google agreed to pay $920 million a month from October 2026 through June 2029 for access to computing capacity. The cost could affect Alphabet's margins when the company reports results on October 28. Alphabet also holds a large long-term investment in SpaceX.
When Meta reports, investors will focus on management's expectations for fiscal 2027 capital expenditure. Meta shares have recently been supported by optimism around its new AI model, “Muse,” and its enterprise-market strategy. The market broadly expects Meta, Amazon, Microsoft, Google and Oracle to reaffirm large-scale AI investment plans. At the same time, investors are assessing whether revenue growth next year can cover that spending and return free cash flow at some companies to positive territory.
Chip stocks retreat as AI performance diverges
Semiconductor stocks have come under pressure. The Philadelphia Semiconductor Index, or SOX, fell 10% last week after more than doubling between late March and June 22. The index delivered its strongest quarterly performance on record in the quarter ended in June, but is now 23% below its record high. Nvidia shares were down 6% through October, while memory-chip companies including Micron fell more sharply amid concerns about Chinese competition and product pricing.
Nvidia will hold its GTC conference in Berlin this week, with Chief Executive Jensen Huang expected to deliver the keynote on Wednesday. AI-related stocks have otherwise posted sharply different returns this year. As of October 16, AMD was up 147% for the year, CrowdStrike had gained 94% and Lumentum had risen 123%. Oracle was down 34% and Broadcom had fallen 5%. In the third quarter, Meta gained 29% and Microsoft rose 37%, while Credo Technology fell 22%, underscoring the market's differing views of growth prospects and investment returns across AI businesses.
Data center optical-network suppliers also came under pressure in October, including Lumentum, Coherent, Credo Technology and Applied Optoelectronics. Arista Networks, Celestica, Akamai Technologies and Hewlett Packard Enterprise also declined. Celestica is scheduled to report results on October 27, with Google having become one of its biggest growth drivers. Investors will use upcoming technology earnings, capital expenditure guidance and compute-contract disclosures to gauge the effect of AI infrastructure expansion on cash flow and supply chains.