Who makes hyperscalers for AI data centers
Hyperscalers, neoclouds & model labs (the tenants and owners)
The hyperscalers, neoclouds, and AI labs that occupy data center capacity as tenants or owner-operators, and whose capital spending ultimately drives demand for every layer below this one.
These are the buyers: the companies that lease or self-build the buildings from Layer 9.01's developers and REITs, fill them with the compute from Layers 1 through 4, and sell the resulting capacity as cloud services or use it to train and run their own models. The category splits into three tiers with different economics — diversified hyperscalers funding data centers from broad cloud and advertising revenue, neoclouds that are effectively single-purpose GPU-leasing vehicles often financed against the hardware itself, and model labs whose compute is frequently backstopped by a hyperscaler partner or dedicated project financing. This layer doesn't manufacture anything, but it's the layer every other part of the taxonomy is downstream of: capital expenditure guidance from this group is the earliest signal for order books across silicon, networking, power, and construction. For investors, the key distinction is balance-sheet quality and revenue diversity — a hyperscaler's data center spend is one line in a much larger business, while a neocloud's spend is the business, which makes its debt and lease commitments far more sensitive to utilization and to the resale value of GPU generations as they age.
AI delta: AI training and inference workloads are the primary driver of the capital spending that funds every other layer of this taxonomy.
Companies on this part (18)
| Company | Role | Ownership | Ticker | Price | 1d | Confidence | Note |
|---|---|---|---|---|---|---|---|
| Alibaba | operator | public | BABA | 99.64 USD | -1.7% | inferred | region: China and Asia |
| Alphabet (Google) | operator | public | GOOGL | 320 USD | -1.3% | inferred | |
| Amazon (AWS) | operator | public | AMZN | 248 USD | -1.1% | inferred | |
| Apple | operator | public | AAPL | 348 USD | -0.3% | inferred | |
| Baidu | operator | public | BIDU | 77.20 USD | -0.7% | inferred | region: China and Asia |
| KT | operator | public | 030200.KS | 46800 KRW | -1.8% | inferred | region: China and Asia |
| Meta Platforms | operator | public | META | 643 USD | -0.5% | inferred | |
| Microsoft | operator | public | MSFT | 525 USD | +0.5% | inferred | |
| Naver | operator | public | 035420.KS | 173600 KRW | -1.3% | inferred | region: China and Asia |
| Oracle | operator | public | ORCL | 126 USD | +0.8% | inferred | |
| Samsung SDS | operator | public | 018260.KS | 213500 KRW | +3.6% | inferred | region: China and Asia |
| SK Telecom | operator | public | 017670.KS | 74500 KRW | -2.1% | inferred | region: China and Asia |
| Tencent | operator | public | 0700.HK | 386 HKD | +0.9% | inferred | region: China and Asia |
| ByteDance | operator | private | inferred | region: China and Asia | |||
| G42 (Khazna) | operator | private | inferred | region: Gulf | |||
| Huawei | operator | private | inferred | region: China and Asia | |||
| Humain | operator | private | inferred | region: Gulf | |||
| Stargate UAE | operator | unknown | inferred | region: Gulf |
Neoclouds 9.02a
| Company | Role | Ownership | Ticker | Price | 1d | Confidence | Note |
|---|---|---|---|---|---|---|---|
| CoreWeave | operator | public | CRWV | 54.23 USD | -0.6% | inferred | |
| DigitalOcean | operator | public | DOCN | 93.22 USD | +1.3% | inferred | |
| IREN | operator | public | IREN | 24.91 USD | -0.9% | inferred | |
| Nebius | operator | public | NBIS | 159 USD | +2.0% | inferred | |
| Cerebras | operator | private | inferred | Cloud | |||
| Crusoe | operator | private | inferred | ||||
| Firmus (Sustainable Metal Cloud) | operator | private | inferred | ||||
| Fluidstack | operator | private | inferred | ||||
| Groq | operator | private | inferred | Cloud | |||
| Lambda | operator | private | inferred | ||||
| Nscale | operator | private | inferred | ||||
| Together | operator | private | inferred | ||||
| Vultr | operator | private | inferred |
Labs 9.02b
Who competes here
The same table read the other way: every company above competes for hyperscalers, neoclouds & model labs sockets. Confidence tags matter — an inferred placement is a lead, not a confirmed rivalry.
Seen in real machines
hyperscalers, neoclouds & model labs is named in 1 of the rack teardowns and supercomputer breakdowns on this site. Each line below is a piece of that machine. The companies shown are the ones named on that line, not necessarily the maker of this part.
- Eagle (Microsoft Azure) · supercomputer
- System design, operation and the ND H100 v5 node (8x H100 GPUs, 96 Xeon cores, 1,900 GiB DDR5, 8x 400 Gb/s InfiniBand adapters per VM; about 1,800 nodes in the TOP500 run) — Microsoft
Nearby parts
- Children: Neoclouds, Labs
- Previous in layer: Data center developers, operators & REITs
- Next in layer: Architecture, engineering & MEP design
- Layer: Layer 9 — Building shell, site & construction
From the blog
Watch
Independent videos — no affiliation; nothing loads from YouTube until you click.
Latest feed
- New Libyan telecom firm to launch data center in 2027 — Data Center Dynamics, 2026-10-23A new Libyan telecom firm plans to launch a data center in 2027, while construction continues on the Medusa subsea cable in Benghazi.
- Meta details Ohio Prometheus data center campus, with first phase live — Data Center Dynamics, 2026-10-23Meta has detailed its Prometheus data center campus in Ohio, with the first phase of the facility already operational.
- Sponsored: Why is Kazakhstan building Central Asia’s first Tier IV data center? — Data Center Dynamics, 2026-10-23Kazakhstan is restructuring its energy market rules, moving away from cryptocurrency mining, to attract high-density artificial intelligence workloads to build a Tier IV data center.
- Microsoft files for data center campus in Floyd County, Georgia — Data Center Dynamics, 2026-10-22Microsoft has filed plans for a data center campus in Floyd County, Georgia, which will consist of three separate buildings.
- Canada's Uniserve to launch 1MW data center in Vancouver — Data Center Dynamics, 2026-10-22Uniserve is launching a one-megawatt data center facility in Vancouver.
- Crusoe signs on for Polar DC's upcoming 40MW data center in Norway — Data Center Dynamics, 2026-10-22Crusoe is securing a deal to build a 40MW data center at Polar's Tørdal campus in Norway.
- Chief Telecom's CTSP AI data center bookings exceed 50% — DigiTimes, 2026-10-21Chief Telecom has started construction on a new AI data center in Central Taiwan Science Park, with over 50% of the facility already booked by international cloud service providers and fintech firms.
- Memphis councilors reject 18-month data center moratorium — Data Center Dynamics, 2026-10-21The Memphis councilors rejected an 18-month moratorium on data centers and instead implemented a 135-day ban on new projects.
- Startup neocloud Aolani signs LOI for 500MW of capacity in Malaysia — Data Center Dynamics, 2026-10-21Neocloud startup Aolani has signed a Letter of Intent (LOI) to secure 500MW of capacity in Malaysia, bringing its total commitment to 580MW.
- SC Capital Partners invests in 40MW data center in Pohang, South Korea — Data Center Dynamics, 2026-10-21SC Capital Partners is investing in a 40MW data center facility in Pohang, South Korea, which will house NeoAI Cloud.
Questions about hyperscalers, neoclouds & model labs
- What are hyperscalers, neoclouds & model labs?
- The hyperscalers, neoclouds, and AI labs that occupy data center capacity as tenants or owner-operators, and whose capital spending ultimately drives demand for every layer below this one.
- Which companies make hyperscalers for AI data centers?
- This catalog lists 18 companies on Hyperscalers, neoclouds & model labs, 13 of them with listed equity: Alibaba, Alphabet (Google), Amazon (AWS), Apple, Baidu, KT, Meta Platforms, Microsoft, Naver, Oracle, and 8 more.
- How does AI change demand for hyperscalers, neoclouds & model labs?
- AI training and inference workloads are the primary driver of the capital spending that funds every other layer of this taxonomy.
- How reliable are these hyperscalers, neoclouds & model labs supplier placements?
- The 18 placements on this part break down as 18 inferred. "disclosed" means the company or a filing says so; "reported" comes from trade press or a reference-design list; "inferred" is a taxonomy lead with no citation yet.
Sources: placements without a source URL come from the taxonomy seed and are tagged accordingly. How this data is built.· Markdown · JSON