From Deflation Engine to Source of Tech Price Inflation
AI infrastructure spending is the rapid, large‑scale investment in chips, data centers, memory and networking gear needed to run and train advanced artificial intelligence systems, and it is powerful enough to reverse decades of falling technology prices by straining component supply, lifting hardware and software costs, and redirecting corporate cash away from shareholder payouts. For roughly twenty years, consumers could count on cheaper laptops, storage, and software, even as performance improved. That pattern is breaking. Data from recent months show computer software and accessories consumer prices rising 14.5% year over year, the largest annual jump since records began in 2000. At the same time, producer prices for electronic components have surged 27% over the same period. Instead of acting as a deflation engine, the tech sector is starting to push inflation higher, with AI data center demand at the center of this shift.
Hyperscaler Capex: A USD 755 Billion AI Arms Race
The most aggressive spending is coming from the hyperscalers, which are racing to build the AI infrastructure layer. Goldman Sachs estimates that the largest AI hyperscalers will spend about USD 755 billion (approx. RM3.48 trillion) on capital expenditures in 2026, an 83% jump from 2025. That figure covers Amazon, Alphabet, Meta, Microsoft and Oracle, whose combined plans range from massive cloud data centers to custom AI chips and robotics. Amazon alone has outlined a roughly USD 200 billion (approx. RM921 billion) capital plan for 2026 tied to AI infrastructure, while Meta has lifted its 2026 capex guidance after citing higher infrastructure and memory costs. Analysts now expect hyperscaler capital expenditures in 2026 to sit near their total cash flows from operations, with a growing role for debt and equity financing. This is AI as industrial buildout, not a side project.
Chip Shortage Economics and Historic Component Price Surges
The capex surge is colliding with finite chip and memory supply, producing classic chip shortage economics and visible tech price inflation. AI data centers consume huge volumes of GPUs, high‑bandwidth memory, DDR4 and DDR5 DRAM, along with storage and networking hardware. Market data show DDR5 and DDR4 memory prices up about 290% year over year, with prices more than doubling in a single year as AI demand absorbs global capacity. At the same time, producer prices for electronic components have risen 27% year over year, while consumer prices for computer software and accessories are up 14.5%. Analysts expect tight conditions in memory and the wider semiconductor supply chain to persist through 2027, supported not only by AI buildouts but also by geopolitical risks. For the first time in two decades, technology hardware and components are seeing sustained price increases instead of annual declines.
Big Tech Buybacks Declining as Hardware Bills Come Due
The AI buildout is also reshaping shareholder returns. Where Big Tech once paired growth with predictable buybacks, rising hyperscaler capex is pulling cash into hardware, data centers and financing costs. Goldman’s analysis shows hyperscaler buybacks falling by nearly two‑thirds in the first quarter, with Microsoft the notable exception among peers. Alphabet bought back no stock in its latest quarter after repurchasing about USD 15.1 billion (approx. RM69.5 billion) a year earlier, while Meta is channeling more cash into infrastructure. According to The Wall Street Journal, free cash flow for the five major hyperscalers is expected to drop 91% in 2026 to about USD 16 billion (approx. RM73.7 billion), even as net income is projected to rise 25% to USD 506 billion (approx. RM2.33 trillion). The old model of Big Tech buybacks declining looks less like a pause and more like a structural shift toward heavy infrastructure investing.
From Growth Stocks to Infrastructure Utilities?
These trends point to a new identity for the biggest AI players. With hyperscaler capex now behaving like that of railroads, utilities or telecom networks, investors are effectively owning infrastructure companies built around AI. The shareholder contract has changed: fewer Big Tech buybacks, more dependence on future AI revenue to justify today’s spending and valuations. Cash that once funded repurchases now competes with data center leases, power contracts, GPUs, memory and rising debt service. Founders partnering with these platforms face a different internal capital allocation process, where every strategic deal must clear a higher financial bar. If enterprise AI adoption and cloud pricing evolve as bulls expect, cash flows could rebound late in the decade. Until then, AI infrastructure spending is both reshaping the economics of chips and storage and ending the long era in which tech products could be relied on to get cheaper every year.
