THE AI LANDSCAPE 04 / INFRASTRUCTURE & COMPUTE

Where intelligence
gets its compute.

Explore the clouds, dedicated clusters, and infrastructure programs behind AI. See who operates them and which capacity is available, planned, or still being brought online.

THE BIG PICTURE

AI Infrastructure

Selected coverage. Check each entry for context and evidence.

8 entries

Each entry links to its primary source ↗

AI Infrastructure. Reviewed 2026-09-23. Editorial classifications, not a performance ranking.
Platform / operatorRoleCompute approachAccess modelCurrent evidence
AWSAmazon1 source Public cloudAI training and inferenceTrainium3Trn3 UltraServersEC2 servicesCustomer cloud accessCurrent product offeringRegions and capacity vary
AzureMicrosoft1 source Public cloudModel and enterprise servicesMaia 200 inferenceCustom accelerator deploymentAzure servicesNot a retail chip offeringMaia deployedJanuary 2026 announcement
Google CloudGoogle1 source Public cloudTPU-based AI computeIronwood / TPU 8Current and next generationsCloud TPUVersion-specific availabilityIronwood GA; TPU 8 soonCurrent catalog status
CoreWeaveCoreWeave1 source Specialist AI cloudGPU infrastructureNVIDIA systemsVera Rubin validationCloud infrastructureCapacity arranged with providerRubin measured on siliconJuly 2026 technical report
Mistral AI CloudMistral AI1 source Regional AI servicesInference and computeEuropean infrastructureRegional endpointsManaged servicesRegional optionsServices + future buildout1 GW ambition by 2030
Meta AI infrastructureMeta2 sources Operator infrastructureMeta AI and productsGPUs + custom MTIAMixed accelerator strategyPrimarily internalNot a general public cloudOngoing deploymentOperator infrastructure update
Colossus 1SpaceX1 source Dedicated clusterContracted computeNVIDIA GPU capacityAnthropic agreementCapacity contractNot a self-service public cloudAgreement announced May 6300 MW cited in agreement
StargateOpenAI + partners1 source Infrastructure programMulti-site buildoutPartner facilitiesPower, sites, and computeOpenAI compute supplyMultiple commercial structuresSecured-capacity milestoneNot all energized capacity
Customer servicesBuildout / delivery milestoneDedicated accessRead the classification notes ↗

A manually reviewed snapshot as of September 23, 2026. Colors organize the evidence; they do not score quality. On smaller screens, swipe across the comparison.

CONTEXT MAKES THE DIFFERENCE

How to read this landscape.

Our interpretation of the comparison.

01

Capacity has stages

Secured power, construction, installed systems, and usable compute are different milestones. Announced gigawatts should not be read as live capacity.

02

A cloud is not just its chips

Networking, storage, orchestration, software, and service access shape the infrastructure a customer actually uses.

03

Ownership and access differ

An exclusive capacity agreement can give a model developer compute without transferring ownership of the data center.

TRANSPARENT BY DESIGN

Follow the evidence.

Official announcements, product documentation, and model repositories. All linked sources were checked on September 23, 2026. Publication dates and review dates are shown separately.

9primary sources
behind this comparison
Suggest a correction
Scope & methodology

Selected infrastructure operators and programs, grouped by their route to use. Cloud services are offered to customers; dedicated compute primarily supports an operator or contracted users; buildout programs aggregate facilities and partnerships over time. Status describes the cited milestone, not a verified site-by-site capacity audit. No market-share or total-capacity ranking is implied.

Dates on continuously updated documentation are labeled “Current documentation” unless a specific publication date is available. This page is maintained manually; the review date is not a live-feed timestamp. Company announcements describe the company’s claims and plans, not an independent audit.

AWS1 source

Amazon / Cloud services

AWS

AWS offers Trn3 UltraServers built around Trainium3, with Neuron software support. Availability of a product family does not mean capacity in every AWS region.

Azure1 source

Microsoft / Cloud services

Azure

Microsoft’s Maia 200 announcement describes deployment inside Azure for inference. This is evidence of an operating custom-chip program, not general customer access to every Maia system.

Google Cloud1 source

Google / Cloud services

Google Cloud

Cloud TPU lists Ironwood as generally available and TPU 8t and 8i as coming soon. The next-generation announcement must not be read as broad customer availability.

CoreWeave1 source

CoreWeave / Cloud services

CoreWeave

CoreWeave reports bringing up and validating Vera Rubin NVL72 and publishing measured results. That establishes a working system, not universal customer availability across its fleet.

Mistral AI Cloud1 source

Mistral AI / Cloud services

Mistral AI Cloud

Mistral’s August update combines regional inference services with a longer-term European compute program. Its target of up to 1 GW by 2030 is a plan, not installed capacity today.

Meta AI infrastructure2 sources

Meta / Dedicated compute

Meta AI infrastructure

Meta describes a network of data centers and a combination of partner accelerators and custom silicon. This row maps the infrastructure role rather than claiming independence from suppliers.

Colossus 11 source

SpaceX / Dedicated compute

Colossus 1

Anthropic announced an agreement for all of Colossus 1’s compute capacity, with more than 300 MW expected within a month. The announcement is evidence of the agreement and schedule, not an independent confirmation of delivered capacity.

Stargate1 source

OpenAI + partners / Buildout programs

Stargate

OpenAI says Stargate surpassed its target of securing 10 GW. Securing infrastructure is different from bringing every facility online; the program spans partners and sites at different stages.

PRIMARY SOURCES

Reviewed 2026-09-23 · AIstify Research