Chips and Gigawatts: How AI Is Doubling the Data Center World by 2030

AI workloads are driving unprecedented growth in both data center chip markets and physical infrastructure, projected to double global data center capacity by 2030. This expansion is critically dependent on overcoming significant power constraints.

Rick Donovan

March 31, 2026 · Updated October 18, 2026 · 5 min read

The data center chip market is on track to nearly triple in value over the next decade, and the buildings that will house those chips are expanding just as fast. Persistence Market Research projects the global data center chip market at US$ 35.2 billion in 2026, rising to US$ 97.1 billion by 2033 at a 15.6% compound annual growth rate. On the physical side, ABI Research forecasts that active data center capacity dedicated to AI workloads will grow from 11.5 GW in 2026 to 43.6 GW in 2031. The two curves are not coincidental: the same AI workload boom that sells silicon is forcing a parallel, power-hungry construction boom.

The AI-Driven Data Center Boom: Chips and Infrastructure

Understanding this market requires reading two ledgers at once. The first is financial: revenue from the processors, memory and networking silicon that make modern data centers run. The second is physical: the gigawatts of electrical capacity needed to switch that silicon on. Both are being reshaped by the same force — the migration of enterprise and cloud workloads toward generative AI and large language models.

The scale of the build-out is unusual. JLL's 2026 Global Data Center Outlook estimates that nearly 100 GW of new data centers will be added between 2026 and 2030, doubling global capacity, and that this expansion will require roughly $3 trillion of investment. That figure makes clear why chip demand and infrastructure demand must be analyzed together: every dollar of accelerator revenue implies a corresponding commitment of land, power and cooling capacity somewhere on the grid.

Explosive Growth in Data Center Chips: A Market Overview

Persistence Market Research's March 2026 report sizes the data center chip market across processors (CPUs, GPUs, FPGAs and ASICs), memory including high-bandwidth memory, and networking components such as interface cards and interconnects. Its headline projection — US$ 35.2 billion in 2026 growing to US$ 97.1 billion by 2033 — implies the market will expand at 15.6% annually across the forecast window.

Within that total, AI accelerators are already the dominant revenue engine. Citing industry research, the report notes that GPUs alone generated US$ 28.5 billion, or 87.6% of a record US$ 32.6 billion in AI data center chipset quarterly revenue in Q4 2024, with quarter-over-quarter growth of 24.8%. That single-quarter figure approaches the size of the entire annual market projection for 2026, illustrating how quickly AI silicon has moved from a niche category to the market's center of gravity.

Parallel growth of the data center chip market and AI data center capacity
Market aspect	2026 figure	Forecast	Primary driver
Data center chip market revenue	US$ 35.2 billion	US$ 97.1 billion by 2033 (15.6% CAGR)	AI and LLM workloads requiring specialized compute
AI-dedicated data center capacity	11.5 GW	43.6 GW by 2031	AI workloads overtaking legacy workloads in active power capacity
Global data center capacity	Baseline year	Nearly 100 GW added by 2030, doubling global capacity	AI and cloud demand; ~$3 trillion investment required
AI share of workloads	About one quarter (2025)	Up to half by 2030, with inference the primary driver	Shift from training-led to inference-led demand

Navigating the Hurdles: Power Constraints and Innovative Solutions

The binding constraint on both curves is electricity. JLL identifies grid constraints as a significant hurdle for the data center build-out, with energy access and power density now strategic considerations rather than afterthoughts. Utility interconnection delays are already changing behavior: some operators are moving beyond power purchase agreements to directly fund their own energy generation.

Regional strategies differ. JLL notes that natural gas is a less prominent solution in EMEA and APAC, where renewables such as solar and wind are seeing increased utilization; in EMEA, projects combining renewables with private wire transmission can reduce tenants' power costs by 40% compared with the grid. Several markets, including Ireland and Texas, have implemented "bring your own power" mandates, which JLL says is fueling the trend toward self-supplied generation. Battery storage and on-site power are correspondingly moving from novelty to standard practice.

These constraints are the practical caveat on every forecast in this article. The 43.6 GW and 97.1 billion dollar figures assume the energy innovations JLL describes arrive on schedule; if grid access lags, chip demand and capacity growth will constrain each other rather than reinforce each other.