Digital Infrastructure
AI Data Centers
AI data centers are specialized digital infrastructure environments designed to support large-scale artificial intelligence training and inference workloads. Combined 2026 capital expenditure from the top hyperscalers (Amazon, Microsoft, Alphabet, Meta, Oracle) is projected at $600-725 billion, up roughly 77% year-over-year, with about 75% dedicated to AI infrastructure and cumulative 2025-2027 hyperscaler capex expected to reach $1.15 trillion.
Sources
Every claim traced to a primary source — evidence, recent activity, and insider filing behavior, all in one place.
Evidence
2 sources · 79% avg. source confidence
- CreditSights / Futurum Group hyperscaler capex analysis — Hyperscaler AI data center capex reaches $600-725B in 2026 across Amazon, Microsoft, Alphabet, Meta, Oracle80% source confidence
- Tom's Hardware / ValueAdd VC hyperscaler capex tracking — Big Four hyperscaler capex hits $725B in 2026, up 77% year-over-year78% source confidence
Dependency Map
What AI Data Centers depends on below, and who depends on AI Data Centers above — click any node to make it the new center, 2 levels deep.
Connection type
Click any node to make it the new center. Scroll to zoom, drag to pan.
The Story So Far (last 6 months)
Auto-generated from this entity's dated milestones, relationship updates, and sourced evidence — not AI-written, just sorted.
Emerging Signals
AI Data-Center Investment Growth
CriticalRapid investment in artificial intelligence infrastructure is increasing demand across semiconductors, memory, networking and supporting physical infrastructure.
Power, Not Chips, Is Becoming the Binding Constraint
CriticalMultiple hyperscalers have publicly cited power and grid interconnection availability, not chip supply, as the primary bottleneck on AI data center growth — Microsoft has disclosed an $80 billion backlog of Azure orders it could not fulfill due to power constraints, driving increased direct investment in nuclear and dedicated power generation.
AI Data Centers's Timeline
A sourced, dated history of AI Data Centers's key moments — founding to present.
Jan 2026 · Hyperscaler capex reaches $600-725B
Combined 2026 capital expenditure from the top hyperscalers (Amazon, Microsoft, Alphabet, Meta, Oracle) reached a projected $600-725 billion, up roughly 77% year-over-year.
How These Connections Evolved
How a relationship changed over time, not just its current state — sourced, dated, and traceable.
AI Data Centers ↔ AI Accelerators
2024–2026
AI data-center buildouts scaled alongside combined hyperscaler capex rising from roughly $226B (2024) to $410B (2025) to a projected $725B (2026), with accelerator supply (NVIDIA GPUs and growing custom-ASIC volumes) remaining the primary gating factor on deployment pace.
Now
AI data centers depend on specialized accelerators for advanced artificial intelligence workloads.
AI Data Centers ↔ NVIDIA
2024–2026
NVIDIA GPUs remained the dominant accelerator platform behind hyperscaler AI data-center buildouts through the Blackwell and Vera Rubin generations, even as custom ASICs gained incremental share.
Now
Many AI data-center deployments depend on NVIDIA accelerated computing platforms.
AI Data Centers ↔ Broadcom
2024–2025
Broadcom's AI networking and custom-ASIC business scaled alongside hyperscaler AI data-center buildouts, becoming a key growth driver as combined capex rose from roughly $226B to $410B.
Now
Advanced data-center infrastructure uses networking and semiconductor technologies supplied by companies including Broadcom.
AI Data Centers ↔ High Bandwidth Memory
2024–2025
HBM demand from AI data-center buildouts drove SK hynix, Micron, and Samsung to ramp HBM3E production and advance HBM4 qualification, with memory supply often gating overall accelerator output.
Now
Expansion of AI data-center infrastructure increases demand for high-bandwidth memory.
AI Data Centers ↔ CoreWeave
Dec 2025
CoreWeave grew to 43 data centers with 850MW+ of active power by the end of 2025, contracting nearly 2GW of additional power during the year for total contracted capacity of 3.1GW+, most coming online by 2027.
Feb 2026
CoreWeave guided to $30-35 billion of 2026 capex aimed at doubling active power capacity past 1.7GW, targeting $12-13 billion revenue and a long-term goal of 8GW+ total active power by 2030.
Now
CoreWeave builds and operates AI data centers as its core business, using this infrastructure category directly.
AI Data Centers ↔ Microsoft Azure
Jan 2026
Microsoft guided to roughly $110-120 billion of 2026 AI data-center capex, among the largest of the big-four hyperscalers, to scale Azure AI infrastructure.
Now
Microsoft Azure operates AI data centers at hyperscale to deliver Azure AI services.
AI Data Centers ↔ Amazon Web Services
Jan 2026
Amazon guided to roughly $200 billion of 2026 AI data-center capex, the largest single commitment among hyperscalers, to scale AWS AI infrastructure including Trainium-based clusters.
Now
AWS operates AI data centers at hyperscale to deliver cloud AI services.
AI Data Centers ↔ InfiniBand (NVIDIA Quantum)
2024–2025
NVIDIA's InfiniBand and evolving Spectrum-X/Quantum-X photonics networking generations became standard for connecting GPU clusters at scale as AI data-center buildouts accelerated.
Now
AI data centers depend on InfiniBand networking to connect GPU clusters across racks at the scale required for large-scale training.
AI Data Centers ↔ Vertiv
2024–2025
Rising GPU rack power density pushed AI data centers toward liquid cooling, with Vertiv commanding roughly 23% global share in precision cooling and power infrastructure for these deployments.
Now
AI data centers depend on liquid cooling and power distribution infrastructure to operate high-density GPU racks that air cooling cannot support.
AI Data Centers ↔ xAI
2024–2026
xAI's Colossus scaled from roughly 100,000-200,000 GPUs in 2024 to a targeted ~555,000 GPUs and ~2GW of power capacity by 2026 with the Colossus 2 expansion.
Now
xAI operates its Colossus supercomputer clusters within large-scale AI data center infrastructure in Memphis, Tennessee.
AI Data Centers ↔ Alphabet (Google)
Jan 2026
Google guided to roughly $175-185 billion of 2026 AI data-center capex to scale both TPU (v7 Ironwood, v8 Sunfish/Zebrafish) and NVIDIA GPU-based compute for its AI services.
Now
Google deploys both TPU and GPU-based compute within large-scale AI data center infrastructure for its AI services.
AI Data Centers ↔ Blackwell
2024–2025
NVIDIA's Blackwell GB200/GB300 systems began deploying into hyperscaler AI data centers, with packaging and memory supply constraints contributing to widely reported ramp delays before production scaled through 2025.
Now
Blackwell systems are designed for deployment in advanced artificial intelligence data-center infrastructure.
AI Data Centers ↔ Micron Technology
2024–2025
Micron ramped HBM3E production and advanced HBM4 qualification to supply AI data-center accelerator deployments, competing with SK hynix and Samsung.
Now
Micron supplies memory technologies used across data-center and artificial intelligence infrastructure.
AI Data Centers ↔ Advanced Micro Devices
2024–2025
AMD's MI300X and MI350X accelerators shipped into AI data centers as a secondary supplier to NVIDIA, with a memory-capacity advantage AMD marketed directly to hyperscalers.
Now
AMD supplies processors and accelerators used in artificial intelligence data centers.
AI Data Centers ↔ AI Accelerators
2024–2026
AI data centers scaled accelerator deployment alongside combined hyperscaler capex rising from roughly $226B (2024) to $410B (2025) to a projected $725B (2026), with the mix increasingly split between merchant GPUs and custom ASICs.
Now
AI accelerators are deployed at scale across artificial intelligence data centers.
AI Data Centers ↔ CoreWeave
2024–2026
CoreWeave scaled from 32 data centers/1.3GW contracted power (2024) to 43 data centers/3.1GW+ contracted (2025), guiding to $30-35B of 2026 capex to double active power capacity.
Now
CoreWeave depends on large-scale data center capacity, power and land to deploy and operate its GPU infrastructure.
AI Data Centers ↔ Microsoft Azure
Jan 2026
Microsoft guided to roughly $110-120 billion of 2026 AI data-center capex to secure the capacity and power needed for Azure's AI infrastructure buildout.
Now
Azure depends on large-scale data center capacity and power availability to deploy AI infrastructure at hyperscale.
AI Data Centers ↔ Amazon Web Services
Jan 2026
Amazon guided to roughly $200 billion of 2026 AI data-center capex, the largest single hyperscaler commitment, to secure capacity and power for AWS AI infrastructure.
Now
AWS depends on large-scale data center capacity and power availability to deploy AI infrastructure at hyperscale.
AI Data Centers ↔ xAI
2024–2026
xAI's Colossus scaled from roughly 100,000-200,000 GPUs (2024) toward a targeted ~555,000 GPUs and ~2GW of power capacity (2026) with the Colossus 2 land acquisition and expansion.
Now
xAI depends on large-scale, power-intensive data center infrastructure to house its Colossus GPU clusters in Memphis, Tennessee.
AI Data Centers ↔ Alphabet (Google)
Jan 2026
Google guided to roughly $175-185 billion of 2026 AI data-center capex to support both custom TPU silicon and NVIDIA GPU-based compute for its AI services.
Now
Google depends on large-scale AI data-center infrastructure to deploy both TPU and GPU-based compute for its AI services.