Advanced Micro Devices (AMD) will sell Anthropic tens of billions of dollars' worth of AI servers, simultaneously investing as much as $5 billion in the Claude maker. This deal strengthens AMD's position in a market largely controlled by rival Nvidia. Anthropic will acquire up to two gigawatts of AMD's latest-generation Instinct MI450 chips, with deliveries beginning in the first half of 2027. AMD's investment in Anthropic is contingent upon the achievement of specific deployment milestones.
AMD executives have stated that one gigawatt of computing power, equivalent to the energy needs of roughly 750,000 U.S. homes, can cost double-digit billions of dollars. This arrangement exemplifies the growing trend of "circular deals" within the AI industry, where chipmakers inject capital into the very AI firms that represent their largest customers. Nvidia, for instance, has engaged in discussions to invest $30 billion in ChatGPT creator OpenAI.
Capital Consolidation Accelerates
Securing AMD chips is critical for Anthropic, providing much-needed AI computing capacity. The startup has rapidly ascended as a leader in Silicon Valley's artificial intelligence race, driven by strong adoption of its enterprise tools like Claude Code. Tom Brown, co-founder and compute head at Anthropic, emphasized the necessity of compute access to maintain Claude's competitive edge and satisfy customer demand. Anthropic has aggressively pursued solutions to overcome capacity constraints for its services in recent months. In May of the same year, it committed to renting the entire computing power of SpaceX's Colossus 1 facility in Memphis, which houses over 220,000 Nvidia processors and added 300 megawatts of new capacity. Meta Platforms is also reportedly in talks to lease computing power to Anthropic, a potential deal valued at up to $10 billion over two years, according to a source familiar with the negotiations last week.
Under the terms of the latest AMD deal, Anthropic plans to deploy some of the chips in its own data centers while leasing additional capacity through existing cloud providers and emerging AI cloud companies, as initially reported by the Wall Street Journal. AMD was also in discussions to provide financial backing for Anthropic's future data-center leases. Shares of the Santa Clara, California-based company saw a 2.4% increase, having more than doubled in value over the current year. In October of the same year, AMD announced a multi-year agreement with OpenAI, projected to generate tens of billions of dollars in annual revenue, while granting the ChatGPT creator an option to acquire approximately 10% of the chipmaker.
The Cost of AI Infrastructure
U.S. hyperscalers are beginning to see returns on their artificial intelligence investments, but the escalating cost of infrastructure buildout is significantly eroding their free cash flow, a trend investors are closely monitoring. At their current trajectory, the five major hyperscalers—Microsoft, Alphabet, Amazon, Meta Platforms, and Oracle—are projected to collectively spend more on capital expenditures than they generate in free cash flow by 2027, according to a Reuters analysis of LSEG consensus estimates. The data indicates these companies will generate approximately $340 billion more in annual operating cash flow in 2027 compared to 2025, yet capital expenditure is expected to surge by roughly $534 billion. This translates to about $1.57 of additional investment for every $1 of additional cash flow.
As these companies report earnings, starting with Alphabet one day ago, investors will scrutinize whether the rapid growth in cloud and AI revenue can keep pace with the anticipated spending surge. Hyperscalers initially led the market rally when the AI buildout commenced, buoyed by promises of future growth. However, over the last year, all but Alphabet have underperformed the S&P 500. Shay Boloor, chief market strategist at Futurum Equities, stated that "Investors are underestimating how fundamentally AI is changing the Big Tech business model." He added that these companies were historically valued as asset-light platforms because revenue could scale much faster than capital requirements, but AI is now pushing them toward a hybrid model where software, advertising, and cloud economics increasingly depend on enormous physical infrastructure spending.
Investors Demand Returns
The capital expenditure estimates encompass all spending, not solely AI-related expenditures, as companies do not consistently disclose AI-specific investments. However, executives confirm that spending on data centers, servers, networking equipment, and other cloud infrastructure is largely driven by AI demand. The spending outlook remains fluid; current-year consensus estimates for the five companies' capital expenditure have climbed from approximately $485 billion in January to around $730 billion in July, according to LSEG. Some indicators suggest AI spending is yielding results: Microsoft reported its AI business has surpassed a $37 billion annual revenue run rate, and Amazon's AWS unit recorded 28% growth in its first quarter.
Despite these figures, investors remain concerned about the potential for AI-related cash generation to falter while spending persists. Microsoft reported $35.8 billion in operating cash flow in its fiscal second quarter, while recording $37.5 billion of capital expenditures, including finance leases. David Russell, global head of market strategy at TradeStation, articulated the core logic of capital: "Earnings growth may not be enough to justify investment if capex is depleting cash. Companies exist to make money, not spend money." Amazon reported a 30% rise in trailing 12-month operating cash flow to $148.5 billion in the first quarter, but its free cash flow simultaneously dropped to $1.2 billion.
Oracle has caused the most alarm among investors, with its shares losing 36% so far this year as its free cash flow has turned negative. Its capital expenditure as a percentage of operating cash flow has steadily increased, rising from 47% in fiscal 2022 to 174% for fiscal 2026, which ended in May of the same year, according to LSEG data. Capital expenditure reached $55.7 billion in its most recent fiscal year, against operating cash flow of $32 billion. Oracle plans to raise $45 billion to $50 billion through debt and equity to fund its cloud infrastructure expansion, a clear move towards debt bondage to sustain growth. Other companies, however, have managed to continue returning cash to shareholders. Microsoft, Alphabet, and Meta still generated sufficient free cash flow to cover dividends and buybacks in their latest fiscal years, according to SEC filings. These buybacks, a direct transfer of wealth to owners, could be jeopardized if spending remains elevated and AI monetization takes longer than anticipated. Freddy Lavric, senior trader at Winthrop Capital Management, stated, "Over the next two to three years, companies need to show that AI is driving incremental revenue, expanding margins and improving cash flow." He concluded that if these financial benefits aren't evident by then, "the market will start questioning whether the investment cycle has gone too far."