Amid market concerns over declining AI capital expenditures, JPMorgan Chase has significantly raised its forecast for data center capex growth, arguing that the market is severely underestimating the earnings potential of AI-related stocks.
According to reports, JPMorgan’s North American equity research team stated in its latest analysis that the arms race among major tech companies is not slowing down but accelerating.
The report projects that data center capex growth in 2026 will exceed 50%, a substantial increase from the previous estimate of 30%. This implies over $150 billion in incremental spending next year, largely covering the expected revenues of core AI chip suppliers such as NVIDIA, AMD, Broadcom, and Marvell.
More importantly, given the persistent and severe lag in AI computing supply relative to demand, JPMorgan believes the actual growth rate could surpass 60%, creating unexpected earnings growth opportunities for AI infrastructure-related firms. The market’s current profit expectations for key players in the AI supply chain remain overly conservative.
**Capex Forecasts Sharply Revised: Explosive Growth Expected in 2026** JPMorgan’s IT hardware team now expects data center spending growth in 2025 to reach around 65%, up from its earlier projection of 55%. This upward revision is driven by hyperscale cloud providers doubling down on AI infrastructure investments, as available AI computing capacity continues to fall far short of demand.
In absolute terms, the incremental spending in 2025 compared to 2024 will exceed $115 billion—significantly higher than the $75–80 billion year-over-year increase in 2024. Notably, this $115 billion increment nearly matches the combined incremental revenue expected this year from NVIDIA, AMD, Broadcom, and Marvell in AI-related GPUs, ASICs, and networking equipment (over $85 billion).
Investors should pay even closer attention to the 2026 outlook. JPMorgan now anticipates data center capex growth exceeding 50% that year, a sharp rise from the prior 30% estimate. At this pace, incremental spending in 2026 would surpass $150 billion.
Historical trends show that upward revisions to capex growth forecasts within a year are common. Both 2024 and 2025 have followed this pattern, with market expectations steadily rising as visibility improves. JPMorgan expects 2026 to repeat this trend, with confidence in sustained growth likely pushing estimates for 2026/2027 even higher.
For the four largest U.S. hyperscale cloud providers, total capex is projected to reach approximately $363 billion in 2025, up ~65% year-over-year, and $447 billion in 2026, maintaining strong momentum.
**Wall Street’s Blind Spot: Underestimated Chip Revenues** For core AI infrastructure suppliers like NVIDIA, Broadcom, AMD, and Marvell, this means consensus estimates are lagging reality. JPMorgan argues that current market revenue projections for these companies in 2026 fail to fully account for the impending $150–175 billion in new capex.
*"Given the robust (and urgent) demand for AI computing, we wouldn’t be surprised if 2026 data center capex growth ultimately lands above 60%... This would unlock AI revenue upside not yet priced into Wall Street forecasts."*
In short, as capex translates into orders, analysts will once again be forced to raise earnings estimates for these chip giants.
**Order Backlogs and Overlooked ‘Non-Traditional’ Buyers** The report highlights how Broadcom and NVIDIA’s backlog values are being misread by the market. For instance, JPMorgan believes investors have misinterpreted Broadcom management’s comments on its "$73 billion 18-month backlog," underestimating how quickly it will convert into revenue.
More critically, market focus remains narrowly fixed on the "Big Four" U.S. cloud providers, while overlooking other rising forces.
*"Limiting the analysis to the top four or five U.S. hyperscalers ignores significant spending occurring outside traditional hyperscale players—including increasingly influential ‘neo-clouds’ and sovereign AI initiatives as they scale."*
This means sovereign nations and emerging cloud platforms are becoming new pillars of AI chip demand alongside tech giants, extending the industry’s growth runway further and stronger than expected.
Comments