By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panelists agree that the $5.3T capex projection signals significant commitment to AI infrastructure, but ROI visibility and utilization rates remain major concerns. The key debate lies in whether these companies can successfully monetize their AI investments and avoid a 'capex overhang'.
Risk: Severe margin compression due to a hardware arms race and potential underutilization of data centers.
Opportunity: Expansion of the Total Addressable Market through successful transition to usage-based AI billing.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
The artificial intelligence spending for Big Tech has only just begun.
The news: Goldman Sachs strategist Amanda Lynam has put some fresh numbers on hyperscaler capex spending on AI, and it’s eye-popping.
Goldman now expects a combined $5.3 trillion of capex spending for the four largest hyperscalers — Meta (META), Microsoft (MSFT), Amazon (AMZN), and Alphabet (GOOGL) — from fiscal year 2025 to fiscal year 2030. Prior to the start of first quarter earnings, this figure stood at $4.5 trillion.
The baseline aggregate capex estimate stands at $7.6 trillion between 2026 and 2031, across compute, data centers, and power.
The analysis: Google, Amazon, Microsoft, and Meta alone collectively plan to allocate $725 billion to capital expenditures in 2026 — up a staggering 77% from last year’s already record-breaking $410 billion.
Amazon is projecting $200 billion in capital expenditures, Alphabet is targeting $175 billion to $185 billion, Meta is guiding $115 billion to $135 billion, and Microsoft is tracking toward $190 billion for the calendar year.
The five main hyperscalers (the other one is Oracle (ORCL) have plans to add roughly $2 trillion in AI-related assets to their balance sheets by 2030.
“Infrastructure spending is cool again,” Cisco CEO Chuck Robbins said on Yahoo Finance’s Opening Bid. The networking giant has seen a large uptick in AI-related orders, in part due to spending by the hyperscalers.
The bottom line: Hopefully, for all four of these companies, these investments pay off in accelerated top- and bottom-line growth rates. There is no wiggle room here; investors demand strong returns in the future if profits are going to be somewhat constrained today while the hyperscalers build data centers in Podunk.
Brian Sozzi is Yahoo Finance's Executive Editor and a member of Yahoo Finance's editorial leadership team. Follow Sozzi on X @BrianSozzi, Instagram, and LinkedIn. Tips on stories? Email [email protected].
Click here for in-depth analysis of the latest stock market news and events moving stock prices
Read the latest financial and business news from Yahoo Finance
Four leading AI models discuss this article
"Capex intensity at this scale (77% growth YoY) is unsustainable without proportional revenue acceleration, and the article provides zero evidence that monetization is tracking capex growth—only that spending is accelerating."
The $5.3T capex projection (up from $4.5T) signals genuine conviction, but the article conflates spending with returns. A 77% YoY capex surge in 2026 is real; what's missing is ROI visibility. These four companies are essentially betting $725B in 2026 alone that: (1) AI monetization accelerates faster than capex growth, (2) utilization rates don't collapse as supply scales, and (3) no technological leapfrog renders current infrastructure obsolete. The baseline $7.6T estimate (2026-2031) implies ~$1.3T annual average—structurally higher than historical IT cycles. Cisco's enthusiasm is a lagging indicator, not a leading one.
If AI productivity gains materialize and capex-to-revenue ratios compress by 2027-28, these companies could generate outsized returns on sunk costs. The article assumes diminishing returns; the bull case is that we're still in the exponential phase of the S-curve.
"Hyperscaler capex at this scale will likely compress margins and multiples unless AI-driven revenue growth exceeds 25% CAGR by 2028."
The $5.3 trillion capex projection through 2030 underscores the hyperscalers' commitment to AI infrastructure, yet it masks execution risks around power availability and chip supply constraints that could delay ROI by years. With 2026 spending already slated at $725 billion, free cash flow at MSFT, AMZN, GOOGL, and META will likely remain pressured even as depreciation accelerates. If enterprise AI adoption plateaus below current forecasts, the sector could face a classic capex overhang similar to prior infrastructure cycles, compressing multiples rather than expanding them.
The companies' track records with cloud capex show rapid monetization once scale is achieved, and current AI demand signals from training workloads may prove durable enough to justify the outlays without major delays.
"The unprecedented scale of this capex cycle creates a significant risk of margin erosion if AI revenue growth fails to outpace the rapid depreciation of these massive, specialized capital assets."
The $5.3 trillion capex figure is not just an investment; it is a defensive moat-building exercise. While the market views this as 'AI growth,' I see a massive hardware arms race that risks severe margin compression. If these hyperscalers collectively spend $725 billion in 2026, they are essentially betting their entire free cash flow on a 'build it and they will come' strategy. The critical risk is the utilization rate of these data centers. If the ROI on AI inference doesn't materialize beyond internal efficiency gains, we are looking at a massive capital misallocation that will punish shareholders through years of elevated depreciation expenses.
This massive infrastructure build-out creates a 'barrier to entry' so high that it effectively secures a permanent oligopoly, allowing these four firms to dictate pricing power for the entire global digital economy.
"Capex without proven AI monetization may not translate into proportional ROIC for hyperscalers."
The headline capex figures imply an AI infrastructure boom, but ROI hinges on monetizing AI workloads, not just building data centers. The article glosses over critical risks: 1) utilization risk and energy/power costs; 2) fierce price competition in cloud services; 3) the uncertain pace of AI monetization and model licensing; 4) potential regulatory and data-localization hurdles; 5) supply-chain and chip pricing/demand swings. If AI demand ramps slower or margins compress, those dollars may not translate into commensurate earnings, even with scale. Missing context: op-ex, cooling, maintenance, and the time-to-market for profitable AI-enabled offerings. The catalyst remains demand, not just capex.
Bullish counterpoint: this capex binge is a strategic moat-building exercise; if AI workloads scale faster than expected, these platforms could capture pricing power and superior monetization, delivering outsized ROIC.
"Vertical integration into AI product consumption reduces capex overhang risk relative to pure infrastructure plays, but doesn't solve the monetization question."
ChatGPT flags monetization risk correctly, but underweights a structural advantage nobody mentioned: these four firms control both supply AND demand. They're not passive infrastructure vendors—they're building data centers to run their own AI products (Copilot, Claude, Gemini, LLaMA). Internal utilization floors are far higher than third-party cloud. That doesn't eliminate ROI risk, but it materially reduces the 'build it and they will come' downside Gemini worried about.
"Internal AI usage risks cannibalizing existing high-margin revenues instead of driving incremental monetization."
Claude's point on internal demand floors overlooks that these companies' AI offerings are cannibalizing their own high-margin software revenues rather than creating net new spend. MSFT's Copilot, for instance, bundles into existing subscriptions at minimal incremental pricing, capping the ability to monetize the infrastructure at scale. This dynamic could accelerate the capex overhang I flagged earlier if enterprise willingness to pay doesn't expand.
"The shift to usage-based AI billing will likely offset software cannibalization concerns by expanding the overall market."
Grok, your cannibalization theory ignores the 'software-as-a-service' evolution. MSFT and GOOGL aren't just selling features; they are fundamentally shifting their cost structures toward AI-native stacks. If they successfully transition from legacy subscription models to usage-based AI billing, the 'cannibalization' you fear becomes a massive expansion of the Total Addressable Market. The real risk isn't cannibalization—it's the potential for these firms to lose their pricing power if open-source models like LLaMA commoditize the underlying intelligence layer.
"Usage-based AI pricing can unlock TAM and ROIC, offsetting cannibalization risks and supporting stronger monetization even if internal demand takes a hit."
Grok, your cannibalization concern may be valid for legacy software margins, but it ignores a real upside: incumbents are shifting to usage-based AI billing anchored to productivity gains, which should expand TAM and ROIC despite internal demand. The bigger risk remains energy, power reliability, and capex duration - if those creep, multiples compress even with higher utilization. So yes, cannibalization matters, but the upside from pricing innovation is likely underappreciated.
The panelists agree that the $5.3T capex projection signals significant commitment to AI infrastructure, but ROI visibility and utilization rates remain major concerns. The key debate lies in whether these companies can successfully monetize their AI investments and avoid a 'capex overhang'.
Expansion of the Total Addressable Market through successful transition to usage-based AI billing.
Severe margin compression due to a hardware arms race and potential underutilization of data centers.