Amazon's Infrastructure Gambit: Satellites and AI Fuel a $200 Billion Bet

Amazon accelerates its AI and satellite broadband push with a major OpenAI partnership and potential Globalstar acquisition, backed by a $200B infrastructure budget.

Amazon is placing a colossal wager on the future of connectivity and artificial intelligence, with a potential satellite acquisition and a blockbuster AI partnership highlighting its aggressive infrastructure push. The tech giant is reportedly in advanced talks to acquire satellite operator Globalstar, a move aimed at accelerating its lagging "Amazon Leo" broadband constellation. Simultaneously, a leaked internal memo from OpenAI reveals overwhelming enterprise demand for its new partnership with Amazon Web Services (AWS), validating the cloud unit's strategic direction.

The reported interest in Globalstar stems from pressing regulatory deadlines. The U.S. Federal Communications Commission (FCC) requires Amazon to have approximately 1,600 satellites in orbit by July 2026. Currently, only 243 are operational, and the company expects to have just 700 ready by this summer. While Amazon has requested an extension to 2028, acquiring Globalstar would provide immediate access to existing infrastructure and valuable frequency licenses in over 120 countries, significantly speeding up network deployment.

Financing this dual-front expansion is not an issue. Amazon has earmarked a staggering $200 billion in planned capital expenditures for 2026, with the bulk dedicated to AI infrastructure. About $1 billion of that is specifically allocated to the Leo satellite program. The market has responded positively to this expansive vision, with the stock gaining roughly 11 percent over the past seven days and trading recently at 204 euros.

The OpenAI partnership, announced in late February, is already proving its worth. An internal memo cited by CNBC reveals that OpenAI's new head of sales, Denise Dresser, described incoming customer interest as "overwhelming." The deal is architecturally distinct: while Microsoft's Azure remains the exclusive provider for stateless APIs, AWS secures rights to sell stateful runtime environments where AI models retain memory and context. OpenAI has committed to spend over $100 billion within this framework.

This structural bet is bolstered by Amazon's in-house chip development. Analysts at Evercore suspect Anthropic's recently unveiled "Mythos" model was trained on Amazon's proprietary Trainium chips. With an annual revenue run rate exceeding $20 billion for its entire chip portfolio, Amazon is embedding itself deeply in the AI infrastructure stack. The company now financially backs two leading independent AI labs—OpenAI and Anthropic—positioning AWS as a foundational layer regardless of which model prevails.

The potential Globalstar acquisition, however, comes with complications. A key stakeholder could complicate negotiations: Apple holds a 20 percent stake in the satellite company. Any final agreement would require complex discussions with the iPhone maker.

The commercial rollout of Amazon's satellite network is already underway. Since April 8, the system has been in a beta phase for government and enterprise customers, achieving data rates of up to 400 Mbps for standard terminals and over 1 Gbps for specialized maritime and aeronautical units. A full public launch of the broadband service is scheduled for the end of 2026. Market observers see significant synergies, particularly for AWS, which would benefit from improved connectivity in remote regions. AWS's AI business already generates annualized revenue of $15 billion.

The financial test for this comprehensive strategy arrives with first-quarter 2026 earnings. Amazon is targeting an operating profit between $16.5 billion and $21.5 billion. Investment bank Citi has already raised its estimates, forecasting 28 percent year-over-year revenue growth for AWS, directly driven by the ramping partnerships with Anthropic and OpenAI.