The demand for AI computing power continues to surge: NVIDIA's supply remains critically low, while Google's TPU leads as ASIC follows closely behind.

JPMorgan's latest CIO survey indicates a substantial increase in AI investment, with 68% of respondents planning to allocate over 5% of their IT budget to AI in the next three years, driving a projected 41% compound annual growth rate in AI-related computing expenditure. This confirms a multi-year build-out of AI infrastructure, supporting strong growth for beneficiaries like NVIDIA, whose market cap Loop Capital forecasts could reach $6 trillion. Concurrently, the report highlights the accelerating importance and market share expansion of AI ASICs, driven by demand, NVIDIA's capacity constraints, and major tech giants' collaborations, suggesting ASICs will increasingly challenge GPUs for AI computing dominance, especially for inference workloads.

Analysis

A recent JPMorgan survey of 168 Chief Information Officers signals a dramatic acceleration in enterprise AI investment, underpinning a multi-year infrastructure build-out. The report indicates the percentage of CIOs allocating over 5% of their IT budget to AI will surge from 25% currently to 68% within three years. This translates to a projected 41% compound annual growth rate in AI-related computing expenditure, outpacing even bullish semiconductor revenue forecasts. While this secular trend supports continued strong growth for GPU leader NVIDIA—bolstered by CEO commentary on exponential demand and a Loop Capital forecast for a potential $6 trillion market cap—it also illuminates a significant secondary trend: the rise of custom AI ASICs. Capacity constraints at NVIDIA, coupled with a push by hyperscalers like Google, Microsoft, and Meta to optimize inference workloads, are driving collaboration with ASIC giants like Broadcom and Marvell. The survey suggests the AI hardware market is shifting from near-total GPU dominance (90% share) towards a more balanced ecosystem, with ASICs poised to capture significant market share. While the report flags potential short-term caution in H2 spending due to trade and geopolitical factors, the long-term outlook across the entire AI compute landscape—both GPU and ASIC—is exceptionally strong, reinforced by the 'Jevons Paradox' where efficiency gains in AI models are expected to spur greater, not less, aggregate demand for computing power.

Key Decisions for Investors

• The strong CIO survey data, projecting a 41% CAGR in enterprise AI spending, validates maintaining a long-term overweight position in the AI hardware ecosystem.
• Investors should consider diversifying beyond a pure-play GPU exposure by increasing allocations to key AI ASIC enablers, such as Broadcom (AVGO), Marvell (MRVL), and Alphabet (GOOGL), to capture the trend of workload-optimized custom silicon.
• Monitor for potential near-term volatility related to flagged geopolitical and trade risks, which could create more attractive entry points for long-term positions in the sector.
• The projected growth in cloud spending to 38% of IT budgets confirms the durable capital expenditure cycle for hyperscalers, supporting positions in Microsoft (MSFT) and Alphabet (GOOGL) as key beneficiaries of the AI build-out.