Global technology giants are turning capital into computing power at an unprecedented pace. Amazon, Google, Microsoft and Meta are set to spend a combined total of more than $700 billion on AI-related capital expenditures in 2026 alone. At the same time, after a deep market correction earlier this year, AI infrastructure valuations have fallen back to their most attractive levels in nearly seven years.
A Structural Shift in Capex
What has unfolded since the start of 2026 is not a series of incremental increases, but a wholesale structural transformation. According to the latest research from TrendForce, driven by substantial upward revisions to capex guidance from North American cloud service providers, total combined capital expenditures for the world’s nine largest CSPs are now projected to reach about $830 billion in 2026 – representing an annual growth rate that has been raised from 61% to 79%.
Amazon alone plans to invest roughly $200 billion in data centre expansion and AI chip deployment, while Microsoft has raised its capex to $190 billion, an increase of about 130% year-over-year. Morgan Stanley estimates that the four big tech giants will spend $630 billion on data centres and AI chips alone.
Gartner chief analyst John-David Lovelock notes in a recent report that the drivers of IT spending growth are now highly concentrated in AI infrastructure, IaaS and generative AI software development. Data centre systems spending is expected to lead with 55.8% growth, approaching a total of $7.88 trillion.
From Capex to Broad-Based Industry Tailwinds
Vast capital outlays are rapidly percolating downstream, and key players across the AI infrastructure supply chain are riding a clear tailwind. Strong demand for high-performance computing clusters from hyperscale data centres continues to drive incremental gains for Nvidia, whose data centre revenue has repeatedly set new quarterly records, maintaining high growth of nearly 75% year-over-year. Broadcom reported AI revenue of $8.4 billion in its first fiscal quarter, up 106% year-over-year, with custom AI accelerators ramping up significantly and companion networking chips beginning to show scale effects in high-bandwidth Ethernet clusters.
At the same time, AI server shipments are expected to grow more than 20% for the full year, with liquid cooling penetration increasing in tandem. Google Cloud saw first-quarter revenue jump 63% year-over-year to $20 billion, while its cloud backlog nearly doubled to $462 billion, and operating margin reached 32.9%. AWS revenue growth also significantly exceeded market expectations. These figures indicate that infrastructure capital spending is translating into quantifiable revenue and profit at a remarkable pace – the fundamental inflection point has already arrived.
Valuation Margin of Safety After the Sharp Pullback
The stock market volatility early this year pushed AI concept stocks sharply lower from historically extreme valuations. For investors accustomed to lofty premiums, current price levels present a new dimension. Morningstar research indicates that the valuation discount in the AI theme sector is now the largest since 2019. Chief equity strategist Michael Field commented: “AI is not a bubble about to burst. The underlying fundamentals are very solid. Semiconductor demand continues to exceed expectations, and core growth drivers like data centres remain intact.”
Although the relentless rise in spending had raised some questions about returns on capital, the long-term logic for AI infrastructure remains undamaged. Goldman Sachs has raised its year-end S&P 500 target to 7,600, arguing that AI investment will contribute about 40% of earnings growth in 2026-2027. JPMorgan has set the same target for the same reason: AI-driven earnings growth is re-emerging as the market’s core engine.
The combination of highly concentrated capital investment and consistently delivered earnings gives the AI infrastructure segment a dual foundation of secular tailwinds and fundamental support. Following the supply chain sequence, subsectors such as chip design, server manufacturing, data centre operations and supporting networking components are poised to capture substantial growth elasticity as capital spending is unleashed.
Risks Remain
The sheer scale of capital spending provides long-term fuel for industry growth, but it is not without headwinds. The fact that some leading CSPs saw their share prices dip temporarily after announcing capex increases reflects potential market concerns about whether such spending can be sustainably converted into solid free cash flow. Meanwhile, GPU supply chain bottlenecks have not been fully resolved, and tariff changes as well as geopolitical factors could also disrupt cross-border equipment supply and increase input costs. Prudent assessment of these risks remains unavoidable when selecting specific names.
The Bottom Line
AI infrastructure is no longer a narrative driven by concept hype. It is a definitive theme backed by undeniable spending data, rapidly materialising industry earnings, and a valuation window that has become increasingly rare after a sharp pullback. 2026 may prove to be the beginning of a much longer expansion cycle.
