Hyperscale Cloud Providers

Snapshot

“Hyperscale cloud providers” describes companies that operate globally distributed compute, storage, networking, and AI infrastructure at extraordinary scale, then sell that capacity as programmable services to enterprises, developers, governments, and increasingly AI labs.
ℹ️
“The worldwide market value for cloud infrastructure is $107 billion in Q3, up 57% from $68 billion eight quarters ago.” [5p5h7d]
This profile captures the category as of October 2026, when traditional hyperscalers are expanding into AI factories while specialized “neocloud” providers are competing through GPU density, faster deployment, and workload-specific economics. The category is worth maintaining as a reference card because AI has made infrastructure capacity, power access, and accelerator supply strategic differentiators rather than merely back-end operating concerns.

What is this Market Category?

Hyperscale cloud providers operate very large data-center fleets and expose their compute, storage, networking, databases, security, developer tools, and AI services through elastic, metered or contracted platforms. Their customers range from startups and software developers to large enterprises, public-sector organizations, and AI model developers that need capacity without owning and operating equivalent infrastructure. The core product is not simply a server rental: it is a globally distributed control plane, an operational platform, and an ecosystem of managed services that lets customers provision and scale workloads programmatically.
The category includes the “big three” public-cloud platforms— Amazon Web Services , Microsoft Azure , and Google Cloud —as well as Oracle Cloud Infrastructure and other large infrastructure platforms with substantial global footprints. These three leading providers accounted for 63% of enterprise cloud-infrastructure spending in one recent quarter, with Amazon at 29%, Microsoft at 20%, and Google at 13%. [5p5h7d] The category increasingly includes specialized AI clouds and “ neoclouds ” such as CoreWeave , Lambda , Crusoe , Nebius , and RunPod when they operate dedicated, large-scale GPU infrastructure rather than merely reselling capacity.
The boundary excludes conventional managed-service providers, colocation operators, enterprise software companies that consume cloud infrastructure, and ordinary hosting firms without hyperscale economics or platform breadth. It also excludes chipmakers such as Nvidia unless they operate or directly finance a cloud service; selling accelerators into the category is adjacent to being a hyperscale provider, not equivalent to providing cloud capacity.
The fuzzy boundary is between a hyperscaler and a neocloud : some operators treat specialized GPU clouds as a new provider class, while others view them as focused infrastructure layers that remain dependent on the major clouds, data-center landlords, chip suppliers, and power markets.

Why Now?

  • AI workloads have converted infrastructure capacity into a strategic bottleneck. Cloud infrastructure revenue reached $107 billion in one reported quarter, while hyperscale capital expenditure rose sharply as providers built AI-oriented capacity. [5p5h7d] [u4yqk5] The enabling shift is that model training and inference require dense accelerator clusters, high-bandwidth networking, and power availability that most enterprises cannot economically assemble themselves.
  • The performance and economics of GPU clouds have crossed a commercial threshold. CoreWeave became the fastest cloud provider to reach $5 billion in annual revenue during 2025, while specialized providers such as Lambda and Crusoe raised billion-dollar-scale financing to build AI factories. [ben15u] This suggests that an infrastructure model once treated as niche GPU rental has become a credible enterprise and AI-lab procurement category.
  • Capital markets are funding infrastructure as a strategic asset class. CoreWeave raised $1.5 billion in its 2025 IPO, while specialized providers attracted multibillion-dollar private financing and debt capacity. [owz493] The capital-formation condition is important: AI-cloud operators need to pre-purchase accelerators, secure power, build facilities, and absorb long periods of utilization ramp before revenue fully catches up.
  • Cloud demand has expanded beyond general-purpose enterprise workloads. Microsoft Azure’s reported growth included a material contribution from AI services, while Google Cloud growth was attributed to core cloud products, AI infrastructure, and generative-AI offerings. [5p5h7d] The major providers can now sell the same underlying platform into application modernization, data analytics, model development, inference, and AI-agent workloads.
  • The “neocloud” concept has given specialized providers a recognizable category identity. Industry coverage increasingly distinguishes GPU-focused providers from general-purpose hyperscalers, with one market analysis estimating that neocloud revenue exceeded $24 billion in 2024 and could reach $65.5 billion by 2030. [owz493] This naming shift matters because it makes a previously fragmented set of GPU hosts legible to investors, cloud buyers, and infrastructure suppliers.

What's Happening?

CAGR and TAM

  • Grand View Research , “Hyperscale Data Center Market Size, Share & Trends Analysis Report, 2025–2030,” estimates the global hyperscale data-center market at $24.5 billion in 2024 , reaching $52.5 billion by 2030 , a 13.6% CAGR from 2025 to 2030 . [a4mhy5] This is a data-center-market estimate, not a complete measure of cloud-service revenue; its framing captures the physical infrastructure required by hyperscale operators.
  • Grand View Research, “Hyperscale Computing Market Size Report, 2023–2030,” estimates a broader hyperscale-computing market of $56.8 billion in 2022 , rising to $130.0 billion in 2026 and $310.1 billion by 2030 , implying a 23.9% CAGR from 2023 to 2030 . [ft6s1i] The materially larger figure reflects a broader boundary that includes computing infrastructure and associated hyperscale systems rather than only hyperscale data centers.
  • The specialized AI-infrastructure segment is growing faster than the conventional hyperscale-data-center market. One forecast values hyperscale AI data-center infrastructure at $22.26 billion in 2025 and projects a 27.5% CAGR from 2026 to 2034 , reaching roughly $176.4 billion by 2034 . [3rnn9v] These figures should be treated as directional because the report’s category boundary is narrower in provider type but broader in AI infrastructure content than traditional cloud-market definitions.
  • The reports disagree because “hyperscale cloud” is not a single standardized market boundary. A physical-facility definition produces a $24.5 billion 2024 base and 13.6% CAGR, while a broader computing definition produces a $56.8 billion 2022 base and 23.9% CAGR. [a4mhy5] [ft6s1i] For strategy work, the relevant distinction is whether the analyst is sizing facilities, cloud services, compute systems, or AI-ready infrastructure.

Category creation events

  • CoreWeave’s IPO crystallized the specialist-AI-cloud model as an investable public category. CoreWeave completed its Nasdaq IPO on March 27, 2025, raising $1.5 billion to scale AI hyperscale-cloud infrastructure. [owz493] The transaction provided a public-market reference point for a provider whose differentiation is accelerator capacity and AI workload specialization rather than the broadest enterprise-service portfolio.
  • The major-cloud providers have made AI infrastructure a primary product battleground. Microsoft Azure’s reported growth included 16 percentage points from AI services in one quarter, while Google Cloud reported growth across AI infrastructure and generative-AI offerings. [5p5h7d] This is a category-creation event in product terms: AI capacity is being packaged as a first-order cloud platform capability rather than an optional workload add-on.
  • The market is coalescing around a two-layer structure: general-purpose hyperscalers and specialized neoclouds. A 2025 mapping of the neocloud landscape estimated more than $24 billion in neocloud revenue in 2024 and projected $65.5 billion by 2030. [owz493] The distinction has become operationally useful even though many specialized providers remain dependent on hyperscaler ecosystems, data-center partners, and accelerator vendors.

Capital concentration

  • Capital is concentrating in providers able to finance accelerator-heavy build-outs. CoreWeave raised $1.5 billion through its IPO, while Lambda raised $480 million in Series D financing and specialized providers pursued further large-scale financing for GPU deployment. [owz493] The financing pattern favors companies that can secure hardware, power, facilities, and anchor customers simultaneously.
  • Nebius has become one of the most visible European AI-cloud financings. Nebius secured $700 million in funding in December 2024 and additional financing in 2025, while expanding across North America, Europe, and the Middle East. [w0gzin] Its positioning combines geographic differentiation with a specialized AI-cloud thesis, including European data-residency and GDPR-oriented demand.
  • The largest concentration of infrastructure spending remains inside the incumbent hyperscalers. Amazon, Microsoft, and Google together represented 63% of enterprise cloud-infrastructure spending in the cited quarter, while their combined infrastructure investment continues to dwarf most specialist-provider budgets. [5p5h7d] Specialized providers therefore compete less by matching the incumbents’ total capital base than by concentrating capital on scarce GPUs, rapid provisioning, and selected workload niches.

Market Incumbents

  • Amazon Web Services — the largest general-purpose hyperscale cloud, with 29% of enterprise cloud-infrastructure spending in the cited quarter and approximately $33 billion in quarterly sales reported for AWS. [5p5h7d]
  • Microsoft Azure — a global hyperscaler integrated with Microsoft’s enterprise software, with Azure and other cloud-services revenue growing 33% in the cited quarter. [5p5h7d]
  • Google Cloud — a hyperscaler differentiated by data, AI infrastructure, and machine-learning services; quarterly revenue reached $15.2 billion in the cited report. [5p5h7d]
  • Oracle Cloud Infrastructure — a large-enterprise cloud provider gaining share through database affinity, AI capacity, and high-growth infrastructure services; Oracle held approximately 3% of the cited cloud-infrastructure market. [5p5h7d]
  • IBM Cloud — an enterprise and regulated-industry cloud focused on hybrid cloud, Red Hat integration, security, and managed infrastructure.
  • Alibaba Cloud — a major China-centered hyperscale platform serving domestic enterprises and international customers, particularly across Asia.
  • NVIDIA — an incumbent accelerator and AI-infrastructure platform company increasingly adjacent to cloud provision through GPU infrastructure, reference architectures, and cloud-provider partnerships.

Amazon Web Services

Stage : public (NASDAQ: AMZN)
Funding : AWS is a business unit of Amazon rather than a separately financed company. Amazon’s cloud segment reported approximately $33 billion in quarterly sales in the cited quarter, with sales up 20% year over year. [5p5h7d]
Footprint : AWS held 29% of enterprise cloud-infrastructure spending in the cited quarter, the largest share among the major providers. [5p5h7d] Its footprint spans global compute, storage, databases, networking, security, analytics, and AI services, although the cited market-share source does not provide a single current region or employee count. [5p5h7d]
Why they're in this category : AWS is the reference general-purpose hyperscaler: it combines the broadest cloud-service catalog with global infrastructure and an enterprise ecosystem, while adding AI infrastructure and managed model services to its established platform. [5p5h7d]
Coverage : The big three grab two-thirds of $107B cloud market in Q3 — TechTarget — reports AWS’s 29% share and quarterly sales growth. [5p5h7d]

Microsoft Azure

Stage : public (NASDAQ: MSFT)
Funding : Azure is a Microsoft business rather than a separately financed company. Microsoft Cloud revenue reached $42.4 billion in the cited quarter, while Azure and other cloud services grew 33% , with AI services contributing 16 percentage points. [5p5h7d]
Footprint : Microsoft Cloud revenue reached $42.4 billion in the cited quarter, and Microsoft held 20% of enterprise cloud-infrastructure spending in that period. [5p5h7d] Azure’s enterprise footprint is reinforced by Microsoft 365, Dynamics, Windows Server, GitHub, and long-standing corporate procurement relationships.
Why they're in this category : Azure’s differentiator is the integration of hyperscale infrastructure with enterprise identity, productivity software, developer tooling, and AI services, allowing Microsoft to convert existing enterprise relationships into cloud and AI consumption. [5p5h7d]
Coverage : The big three grab two-thirds of $107B cloud market in Q3 — TechTarget — details Azure growth and the contribution of AI services. [5p5h7d]

Google Cloud

Stage : public (NASDAQ: GOOGL)
Funding : Google Cloud is part of Alphabet and is not separately financed. The cited quarter’s Google Cloud revenue was $15.2 billion , up 34% year over year. [5p5h7d]
Footprint : Google Cloud held 13% of enterprise cloud-infrastructure spending in the cited quarter and reported $15.2 billion in quarterly revenue in that period. [5p5h7d] Its footprint combines global cloud regions with Google’s data, machine-learning, networking, and accelerator capabilities.
Why they're in this category : Google Cloud competes through AI infrastructure, machine-learning platforms, data analytics, and proprietary accelerator systems, positioning the cloud as the commercial extension of Google’s internal AI and distributed-systems expertise. [5p5h7d]
Coverage : The big three grab two-thirds of $107B cloud market in Q3 — TechTarget — reports Google Cloud revenue growth led by AI infrastructure and generative AI. [5p5h7d]

Market Challengers

  • CoreWeave — public AI-cloud specialist and the clearest challenger to the general-purpose hyperscalers in GPU-intensive workloads; it raised $1.5 billion in its 2025 IPO. [owz493]
  • Nebius — publicly traded or public-market-linked European AI-cloud challenger with substantial 2024–2025 financing and a cross-region expansion strategy. [w0gzin]
  • Lambda — GPU-cloud scale-up deploying large accelerator fleets for AI researchers, developers, and enterprises; it raised $480 million in Series D financing. [owz493]
  • Crusoe — AI-infrastructure scale-up combining data-center development with energy-oriented infrastructure and large GPU deployments; it raised $1.375 billion in Series E financing. [ben15u]
  • IREN — publicly traded data-center and compute operator expanding from digital infrastructure into AI-cloud and GPU capacity.
  • DigitalOcean — developer-focused public cloud challenger with a simpler product experience and growing AI infrastructure ambitions.
  • RunPod — fast-growing GPU-cloud challenger serving developers and AI teams through accessible on-demand accelerator infrastructure.

CoreWeave

Stage : public (NASDAQ: CRWV; IPO in 2025)
Funding : CoreWeave raised $1.5 billion through its March 2025 IPO to fund expansion of its AI hyperscale-cloud infrastructure. [owz493] The company’s financing history also includes substantial private capital and debt facilities, but the cited search results do not provide a complete, audited total raised.
Footprint : CoreWeave became the fastest cloud provider to reach $5 billion in annual revenue during 2025 , according to the cited industry analysis. [ben15u] Another market analysis reported more than $2.1 billion in revenue in the first half of 2025 . [owz493] CoreWeave’s footprint is concentrated in GPU-intensive AI workloads rather than the full breadth of general-purpose cloud services.
Why they're in this category : CoreWeave is the clearest specialist challenger because it commercializes dense NVIDIA GPU capacity as a cloud platform and has converted AI-lab and hyperscaler demand into public-company-scale revenue. [ben15u] [owz493]
Coverage : Mapping the Neocloud Landscape — industry landscape report — identifies CoreWeave as the current leader among specialized neoclouds. [owz493]

Nebius

Stage : scale-up / public-market-linked AI-cloud provider
Funding : Nebius secured $700 million in funding in December 2024 and additional financing of approximately $1 billion in 2025 , according to industry coverage. [w0gzin] [52z90m] The cited results do not identify a single lead investor for each financing event.
Footprint : Nebius reported 479% revenue growth during 2025 and began expanding commercial operations into Asia-Pacific during 2026. [ben15u] It is headquartered in Amsterdam and has been described as building a presence across North America, Europe, and the Middle East. [w0gzin]
Why they're in this category : Nebius is a challenger because it combines specialized AI-cloud infrastructure with a European operating and compliance position, targeting customers that need GPU capacity with geographic or data-residency differentiation. [w0gzin] [52z90m]
Coverage : The rise of Neocloud in the AI landscape — JLL — highlights Nebius’s $700 million 2024 financing and additional 2025 capital. [mzzsg4]

Lambda

Stage : late-stage private scale-up
Funding : Lambda raised $480 million in Series D financing in February 2025 at a reported $2.5 billion valuation , with Andra Capital and SGW identified as lead investors and participation from Nvidia and others. [0s2gou] [owz493] The cited results also report a further $1.5 billion financing in November 2025 tied to a multibillion-dollar Microsoft deal. [ben15u]
Footprint : Lambda’s disclosed operating footprint is centered on deploying GPU capacity, including a reported plan to deploy 65,000 NVIDIA H100 GPUs across data centers. [owz493] The company’s customer and employee counts are not available in the cited results.
Why they're in this category : Lambda is a challenger because it has moved from specialist GPU hosting toward large-scale AI-cloud infrastructure and strategic hyperscaler relationships, while remaining more focused than AWS, Azure, or Google Cloud. [0s2gou] [owz493]
Coverage : Mapping the Neocloud Landscape — industry landscape report — covers Lambda’s Series D, GPU deployment plan, and Nvidia participation. [owz493]

Market Innovators

  • Together AI — developer-oriented AI-cloud startup focused on open models, inference, fine-tuning, and accessible GPU infrastructure.
  • Crusoe Energy — although its later financing places it structurally closer to challenger status, its energy-first AI-factory thesis began as an innovative alternative to conventional data-center development.
  • Vast.ai — marketplace-style GPU cloud that aggregates distributed supply rather than relying exclusively on an owned hyperscale fleet.
  • Modal — developer infrastructure startup abstracting GPU deployment and serverless execution for AI workloads.
  • Baseten — AI-inference infrastructure startup focused on deploying and operating production models for engineering teams.
  • Replicate — model-deployment platform that packages inference access and model execution for developers.
  • Gensyn — novel decentralized-compute thesis attempting to coordinate machine-learning workloads across distributed resources.
The tier classification is inherently dynamic: Crusoe has already raised capital large enough to resemble a challenger, while Together AI, Vast.ai, Modal, Baseten, Replicate, and Gensyn are best treated as innovators when the relevant distinction is novel architecture or developer abstraction rather than absolute funding scale.

Together AI

Stage : late Series B / early scale-up
Funding : The cited search results do not provide a sufficiently reliable current figure for Together AI’s total funding, most recent round, or lead investor. That information should be verified against the company’s financing announcement or a current financial database before investment use.
Footprint : Together AI is positioned around open-source model hosting, fine-tuning, inference, and GPU-backed AI development, but the cited results do not provide a verified current revenue, employee, or customer figure.
Why they're in this category : Together AI belongs in the innovation tier because it attacks the cloud layer through open-model infrastructure and developer-oriented AI services rather than attempting to reproduce the entire general-purpose cloud stack.
Coverage : The available search results mention Together AI as part of the specialist-provider landscape but do not provide a sufficiently specific article title or verified financing detail for a stronger citation. [3rlg0m]
Stage : early-stage startup
Funding : The available search results do not provide a verified total raised, most recent round, lead investor, or round date for Modal. A financing card should therefore be completed from a primary company announcement or current financial database before external circulation.
Footprint : Modal’s publicly visible position is as a developer platform for running serverless and GPU workloads, but the cited results do not provide verified current revenue, employee count, customer count, or geographic reach.
Why they're in this category : Modal represents an innovation thesis that cloud users should consume accelerators through higher-level, serverless execution abstractions rather than provision and manage virtual machines or entire GPU clusters.
Coverage : No sufficiently specific, verified article from the returned search results was available for citation.

Vast.ai

Stage : early-stage / growth startup
Funding : The returned search results do not provide a verified financing total, latest round, lead investor, or date for Vast.ai.
Footprint : Vast.ai is identified in the available industry coverage as part of the long tail of GPU-rental and specialist infrastructure providers. [3rlg0m] The cited results do not provide an audited revenue figure, employee count, or customer number.
Why they're in this category : Vast.ai’s innovation is a marketplace-oriented supply model that can aggregate underutilized GPU capacity across providers instead of depending solely on a centrally owned hyperscale fleet. [3rlg0m]
Coverage : GPU Rental Market Size, Share and Forecast to 2034 — Dataintelo — places Vast.ai alongside CoreWeave, Nebius, Lambda, Crusoe, RunPod, DigitalOcean, and Together AI in the specialist GPU-provider landscape. [3rlg0m]

Industry Coverage and Market Data

Market Reports

  • Hyperscale Data Center Market Size, Share & Trends Analysis Report, 2025–2030 — Grand View Research — estimates a $24.5 billion 2024 market , reaching $52.5 billion by 2030 at a 13.6% CAGR from 2025 to 2030 ; the framing is the global physical hyperscale-data-center market. [a4mhy5]
  • Hyperscale Computing Market Size Report, 2023–2030 — Grand View Research — uses a broader hyperscale-computing boundary, estimating $56.8 billion in 2022 , $130.0 billion in 2026 , and $310.1 billion in 2030 , at a 23.9% CAGR from 2023 to 2030 . [ft6s1i]
  • Hyperscale AI Data Center Infrastructure Market Research Report — MarketIntelo — estimates $22.26 billion in 2025 and forecasts a 27.5% CAGR from 2026 to 2034 , reaching approximately $176.4 billion ; the report focuses on AI-specific hyperscale infrastructure. [3rnn9v]
  • AI-Ready Hyperscale Data Center Market — MarketIntelo — estimates a $205.48 billion 2025 market and forecasts $1.43 trillion by 2034 at a 23.7% CAGR ; this is a much broader AI-ready infrastructure framing and should not be compared directly with a cloud-services TAM. [9mw3oo]
  • Neocloud Market Size, Share, Trends — Market.us — reports a neocloud-growth framing with a 52.6% CAGR , identifies CoreWeave as the largest neocloud provider by the cited period, and highlights Nebius’s 2024 and 2025 financing. [52z90m]

Industry Articles

  • The big three grab two-thirds of $107B cloud market in Q3 — TechTarget — reports Synergy Research Group’s quarterly market-share estimate, including the 63% combined share of Amazon, Microsoft, and Google. [5p5h7d]
  • The rise of Neocloud in the AI landscape — JLL — frames neoclouds as a growing AI-infrastructure layer and highlights CoreWeave’s post-IPO performance, Nebius financing, and sector capital allocation. [mzzsg4]
  • What Is a Neocloud? Why AI Chipmakers Are Becoming Cloud Providers — Michael Hensel — explains the neocloud thesis and cites Nebius’s $700 million financing and global expansion. [w0gzin]
  • 10 Fastest Growing Cloud Infrastructure Companies to Watch — Landbase — highlights the rapid revenue and financing growth of CoreWeave, Crusoe, Lambda, and Nebius. [ben15u]
  • Neocloud Business Model and Unit Economics — American Compute — examines how GPU-cloud companies finance hardware and cites Lambda’s Series D valuation and subsequent strategic financing. [0s2gou]

Financial News Sources

  • The big three grab two-thirds of $107B cloud market in Q3 — TechTarget — recaps cloud-company earnings and Synergy’s market-share data, including AWS, Azure, Google Cloud, and Oracle. [5p5h7d]
  • Mapping the Neocloud Landscape — industry landscape report — reports CoreWeave’s 2025 IPO, first-half revenue, Lambda’s Series D, and the estimated neocloud market trajectory. [owz493]
  • 10 Fastest Growing Cloud Infrastructure Companies to Watch — Landbase — reports CoreWeave’s $5 billion annual-revenue milestone, Crusoe’s $1.375 billion Series E, and Lambda’s $1.5 billion financing. [ben15u]
  • GPU Rental Market Size, Share and Forecast to 2034 — Dataintelo — identifies the specialist-provider set and cites CoreWeave’s reported 2025 revenue, Nebius revenue, and the long tail of GPU-cloud providers. [3rlg0m]
  • Neocloud Market Size, Share, Trends — Market.us — reports CoreWeave’s quarterly revenue, funding, customer references, and Nebius’s financing history. [52z90m]

Frontier and Open Questions

  • Will specialized neoclouds become durable independent hyperscalers, or will the major clouds absorb their economics? CoreWeave, Lambda, Nebius, and Crusoe are most likely to resolve this through GPU availability, utilization, pricing, and anchor-customer retention. [ben15u] [owz493]
  • Can a specialist provider achieve attractive returns when accelerator depreciation, power, networking, and financing costs are included? Lambda and Crusoe are the most important operators to watch because their large financing rounds increase both deployment capacity and balance-sheet exposure. [ben15u] [0s2gou]
  • Will AI infrastructure remain a distinct cloud category or collapse into the product catalogs of AWS, Azure, and Google Cloud? The incumbents’ AI-linked growth and dominant combined market share favor consolidation, while CoreWeave’s IPO and the neocloud funding cycle support continued category independence. [5p5h7d] [owz493]
  • Should “hyperscale cloud” include energy-first AI factories and colocation-backed GPU operators? Crusoe and IREN sit directly on this boundary: their infrastructure may be strategically equivalent to cloud capacity for AI workloads even when their operating model differs from a conventional global cloud platform. [ben15u] [3rlg0m]
  • Will cloud buyers optimize for global service breadth or for specialized performance-per-dollar? AWS, Azure, and Google Cloud have the broader platforms, while CoreWeave, Lambda, Nebius, and RunPod compete through concentrated GPU capacity and workload-specific economics. [5p5h7d] [owz493]
  • Can Europe and other regions sustain differentiated AI clouds based on sovereignty, data residency, and local power access? Nebius is the clearest named operator testing this thesis through European roots and expansion across North America, Europe, the Middle East, and Asia-Pacific. [ben15u] [w0gzin]

Adjacent Concepts and Categories

  • AI Factories — purpose-built facilities that combine accelerators, networking, power, cooling, and orchestration for AI training and inference.
  • AI-Factories-Datacenters — the physical infrastructure layer where hyperscale and neocloud capacity is deployed.
  • Neoclouds — specialized cloud providers focused on GPU-heavy AI workloads rather than the full general-purpose cloud stack.
  • GPU-as-a-Service — on-demand access to accelerator capacity through cloud, marketplace, or dedicated-cluster models.
  • Cloud Infrastructure Services — the broader IaaS, PaaS, and hosted-private-cloud market in which hyperscalers compete.
  • Data-Center Power and Cooling — the energy, grid, liquid-cooling, and facility constraints increasingly determining AI-cloud expansion.
  • Sovereign Cloud — regionally controlled cloud infrastructure designed around jurisdiction, data residency, and regulatory requirements.
  • AI Inference Infrastructure — the serving, optimization, and orchestration layer that turns trained models into production applications.

Sources