In the latest wave of tech earnings, a pronounced surge in capital expenditures related to artificial intelligence (AI) has captured attention, particularly among major players in the industry. Meta and Microsoft have both revised their projections, indicating that their capital expenditure growth in 2026 will outpace that of 2025. Alphabet has also expanded its capital spending plans for the remainder of this year, showcasing the relentless push towards AI.
However, amidst this flurry of investment, a closer inspection reveals a complex landscape marked by varying levels of risk and potential returns. As the fervor for AI-driven growth begins to stabilize, investors are likely to turn their focus toward discerning which companies possess sustainable capital spending practices.
One notable distinction arises from the differing degrees of immediate demand for AI resources. Microsoft, for example, recently extended its forecasts, now expecting that demand for its AI capabilities will outstrip supply until at least mid-2024. The company’s executives highlighted that the contracts already in place align closely with the projected lifespan of its AI chips, facilitating a congruence between costs and revenue.
Conversely, Meta appears to be engaging in more speculative investments. CEO Mark Zuckerberg faced considerable scrutiny during a recent earnings call regarding the rationale behind the company’s capital expenditures. He presented a broad vision of developing new services with mass appeal but did not clarify specific future offerings. Although investors have generally shown confidence in Zuckerberg’s leadership—evident from the substantial recovery of Meta’s stock price from previous lows—there was a notable 10% drop in shares following the earnings report, indicating waning patience among shareholders.
Another relevant factor is the nature of reported AI demand, particularly the backlog of contracted business, known as remaining performance obligations (RPO). Oracle’s recent announcement of a jump to $455 billion in RPO initially buoyed its stock. However, it later became clear that a significant portion—$300 billion—was tied to a long-term deal with OpenAI, which may not generate immediate demand.
Microsoft disclosed that its RPO had increased by 50% to nearly $400 billion, not including a $250 billion agreement with OpenAI. The company emphasized that its bookings derive from diverse customers and feature a relatively short average contract term of two years, suggesting these agreements are likely to produce revenue swiftly. This creates two distinctive AI capital expenditure trajectories: one short-term, grounded in tangible demand, and the other long-term and speculative, rooted in a more visionary belief in growth potential.
Profitability remains a third critical element as data centers increasingly fill with AI servers and other ephemeral assets, leading to rising depreciation costs. Microsoft reported that its depreciation and amortization rose to 16.8% of revenue, up from 11.3% a year earlier. As competition heightens, maintaining expense discipline will be imperative. Cost-effective AI infrastructure and sustainable pricing power will likely emerge as essential characteristics of leading firms in the space.
In the latest quarter, both Google and Microsoft surpassed profit margin expectations, whereas Meta experienced a decline in its operating margin. The ability to defend these margins will be a focal point in 2026. Google is heavily invested in its comprehensive suite of in-house AI technology, while Microsoft aims to avoid the lower-margin segments of the AI market, including the rental of GPU capacity. Despite some investor concerns regarding Microsoft’s shifting relationship with OpenAI, CEO Satya Nadella reassured stakeholders that more lucrative opportunities lie ahead.
As the current enthusiasm for AI investments begins to wane, the strategic decisions being made today will play a pivotal role in determining which companies will prosper in the long term.
