2026-10-11 16:38 UTC

CNBC reports that Anthropic and OpenAI are exploring 20–30 MW compute deals in Europe and the US alongside larger campuses, potentially accelerating usable inference capacity through smaller deployments rather than replacing their megaproject commitments.

state: seedheat: mediumuncertainty: mediumnovelscott: lowai-infrastructure data-centers inference-capacityAnthropicOpenAIKai Nicol-Schwarz

What is this?

The case concerns a reported search by AI model providers Anthropic and OpenAI for smaller data-center capacity deals alongside larger infrastructure commitments. The supplied CNBC snippets establish that Anthropic is pursuing European capacity and evaluating deals directly with developers worldwide; Anthropic’s own announcement also confirms a multi-gigawatt Google/Broadcom agreement expected to begin coming online in 2027. However, the retrieved snippets do not establish the specific 20–30 MW deals, the U.K./Nordics locations, or OpenAI’s participation in that smaller-deal search. Faster delivery of usable inference capacity, rather than replacement of megaprojects, remains the case’s interpretation rather than a demonstrated outcome.

Why it matters to Scott

Scott’s LiteLLM gateway makes provider availability an operational concern, but the supplied material establishes no change to API capacity, pricing or reliability that would alter his routing or agent budgets. No hit shows Scott holding this smaller-site procurement thesis or the radar tracking this same development; the specific deal sizes, OpenAI participation and faster-delivery interpretation remain unestablished by the grounding.
dev:technology.litellmradar:concept.ai-infrastructureradar:concept.datacenters
queries asked of Scott's wikis
  • compute bottlenecks inference availability agent workloads
  • modular infrastructure time to capacity megaprojects
  • inference economics token budgets agent scaling
  • model provider capacity limits multi-provider routing
  • European compute data residency model sovereignty

Measured heat

now 0 pts/hpeak 0 pts/hcomments 0/hpeers p14momentum: steady2 platformsage 578h
points/hour across evidence · reading as of 2026-10-12 02:59:37.977291+11:00 · deterministic, not a model opinion

How the heat travelled

09-17 14:00⭐ origin echo-reconstructedCNBC’s original report says Anthropic and OpenAI are seeking 20–30 MW data-center deals in the U.K., Nordics, and potentially the U.S., base
Kai Nicol-Schwarz (CNBC) on other (echo) · attributed from hn.story.49752586
—
09-18 10:58first on hacker news · published · +21.0hAnthropic and OpenAI hunt for smaller data center deals
cebert
—
09-18 10:58amplified on hacker news 👑hn.story.49752586
cebert
peak 8 · 1 comments · 101% of case engagement
09-18 11:21our radar first saw it · +21.4hdiscovery anchor: hn.story.49752586—
pace: p42 vs 1032 stories at the 336h mark (now 578h old) — ahead of agent-memory-add-search-evaluation (1.2x), behind agenticos-self-hosted-governance (0.9x)

Evidence (2) — ⭐ canonical anchor

sourceobjectauthorscorecomments
🟧 hnAnthropic and OpenAI hunt for smaller data center deals
Retrieved article excerpt

Open article · Retrieved 2026-09-18T11:22:22.902546+00:00

[Skip Navigation](https://www.cnbc.com/2026/09/18/anthropic-openai-small-ai-data-center-deals.html#MainContent)

[CNBC](https://www.cnbc.com/world/)

[Markets](https://www.cnbc.com/markets/)



- [Pre-Markets](https://www.cnbc.com/pre-markets/)
- [U.S. Markets](https://www.cnbc.com/us-markets/)
- [Europe Markets](https://www.cnbc.com/markets-europe/)
- [China Markets](https://www.cnbc.com/china-markets/)
- [Asia Markets](https://www.cnbc.com/markets-asia-pacific/)
- [World Markets](https://www.cnbc.com/world-markets/)
- [Currencies](https://www.cnbc.com/currencies/)
- [Prediction Markets](https://www.cnbc.com/prediction-markets/)
- [Cryptocurrency](https://www.cnbc.com/cryptocurrency/)
- [Futures & Commodities](https://www.cnbc.com/futures-and-commodities/)
- [Bonds](https://www.cnbc.com/bonds/)
- [Funds & ETFs](https://www.cnbc.com/funds-and-etfs/)

[Business](https://www.cnbc.com/business/)



- [Economy](https://www.cnbc.com/economy/)
- [Finance](https://www.cnbc.com/finance/)
- [Health & Science](https://www.cnbc.com/health-and-science/)
- [Media](https://www.cnbc.com/media/)
- [Real Estate](https://www.cnbc.com/real-estate/)
- [Energy](https://www.cnbc.com/energy/)
- [Climate](https://www.cnbc.com/climate/)
- [Transportation](https://www.cnbc.com/transportation/)
- [Investigations](https://www.cnbc.com/cnbc-investigations/)
- [Industrials](https://www.cnbc.com/industrials/)
- [Retail](https://www.cnbc.com/retail/)
- [Wealth](https://www.cnbc.com/wealth/)
- [Sports](https://www.cnbc.com/sports/)
- [Life](https://www.cnbc.com/life/)
- [Small Business](https://www.cnbc.com/small-business/)

[Investing](https://www.cnbc.com/investing/)



- [Personal Finance](https://www.cnbc.com/personal-finance/)
- [Fintech](https://www.cnbc.com/fintech/)
- [Financial Advisors](https://www.cnbc.com/financial-advisors/)
- [Options Action](https://www.cnbc.com/options-action/)
- [ETF Street](https://www.cnbc.com/etf-street/)
- [Buffett Archive](https://buffett.cnbc.com)
- [Earnings](https://www.cnbc.com/earnings/)
- [Trader Talk](https://www.cnbc.com/trader-talk/)

[Tech](https://www.cnbc.com/technology/)



- [Cybersecurity](https://www.cnbc.com/cybersecurity/)
- [AI](https://www.cnbc.com/ai-artificial-intelligence/)
- [Enterprise](https://www.cnbc.com/enterprise/)
- [Internet](https://www.cnbc.com/internet/)
- [Media](https://www.cnbc.com/media/)
- [Mobile](https://www.cnbc.com/mobile/)
- [Social Media](https://www.cnbc.com/social-media/)
- [CNBC Disruptor 50](https://www.cnbc.com/cnbc-disruptors/)
- [Tech Guide](https://www.cnbc.com/tech-guide/)

[Politics & Policy](https://www.cnbc.com/politics/)



- [White House](https://www.cnbc.com/white-house/)
- [Policy](https://www.cnbc.com/policy/)
- [Defense](https://www.cnbc.com/defense/)
- [Congress](https://www.cnbc.com/congress/)
- [Expanding Opportunity](https://www.cnbc.com/expanding-opportunity/)
- [Europe Politics](https://www.cnbc.com/europe-politics/)
- [China Politics](https://www.cnbc.com/china-politics/)
- [Asia Politics](https://www.cnbc.com/asia-politics/)
- [World Politics](https://www.cnbc.com/world-politics/)

[Video](https://www.cnbc.com/tv/)



- [Latest Video](https://www.cnbc.com/latest-video/)
- [Full Episodes](https://www.cnbc.com/live-tv/full-episodes/)
- [Livestream](https://www.cnbc.com/live-tv/)
- [Top Video](https://www.cnbc.com/top-video/)
- [Live Audio](https://www.cnbc.com/live-audio/)
- [Europe TV](https://www.cnbc.com/europe-television/)
- [Asia TV](https://www.cnbc.com/asia-business-day/)
- [CNBC Podcasts](https://www.cnbc.com/podcast/)
- [CEO Interviews](https://www.cnbc.com/video-ceo-interviews/)
- [Digital Originals](https://www.cnbc.com/digital-original/)

[Watchlist](https://www.cnbc.com/watchlist/)

[Investing Club](https://www.cnbc.com/investingclub/subscribe?__source=investingclub|globalnav|join&tpcc=investingclub|globalnav|join)



- [Trust Portfolio](https://www.cnbc.com/investingclub/charitable-trust/)
- [Analysis](https://www.cnbc.com/investingclub/analysis/)
- [Trade Alerts](https://www.cnbc.com/investingclub/trade-alerts/)
- [Meeting Videos](https://www.cnbc.com/investingclub/video/)
- [Homestretch](https://www.cnbc.com/investingclub/homestretch/)
- [Jim's Columns](https://www.cnbc.com/investingclub/jim-cramer-columns/)
- [Education](https://www.cnbc.com/investingclub/education/)
- [Subscribe](https://www.cnbc.com/investingclub/subscribe?__source=investingclub|dropdownnav&tpcc=investingclub|dropdownnav)

Join IC

[PRO](https://www.cnbc.com/application/pro?__source=pro|globalnav|join&tpcc=pro|globalnav|join)



- [Pro News](https://www.cnbc.com/pro/news/)
- [Josh Brown](https://www.cnbc.com/josh-brown/)
- [Mike Santoli](https://www.cnbc.com/mike-santoli-insight/)
- [Calls of the Day](https://www.cnbc.com/analyst-calls-of-the-day/)
- [My Portfolio](https://www.cnbc.com/my-portfolio/)
- [Livestream](https://www.cnbc.com/live-tv/)
- [Full Episodes](https://www.cnbc.com/live-tv/full-episodes/)
- [Stock Screener](https://www.cnbc.com/pro-stock-screener/)
- [Market Forecast](https://www.cnbc.com/2023/05/11/market-strategist-survey-forecast.html)
- [Options Investing](https://www.cnbc.com/pro/options-investing/)
- [Chart Investing](https://www.cnbc.com/cnbc-pro-chart-investing/)
- [Subscribe](https://www.cnbc.com/application/pro/?__source=pro|dropdownnav&tpcc=pro|dropdownnav)

Join Pro

[Livestream](https://www.cnbc.com/live-tv/)

Menu

- [Make It](https://www.cnbc.com/make-it/ "makeit")
- select

- [USA](https://www.cnbc.com/?region=usa)
- [INTL](https://www.cnbc.com/world/)

- [Livestream](https://www.cnbc.com/live-tv/)

Search quotes, news & videos

- [Livestream](https://www.cnbc.com/live-tv/)

[Watchlist](https://www.cnbc.com/watchlist/)

[SIGN IN](https://www.cnbc.com/2026/09/18/anthropic-openai-small-ai-data-center-deals.html)

[Create free account](https://www.cnbc.com/2026/09/18/anthropic-openai-small-ai-data-center-deals.html)

[Markets](https://www.cnbc.com/markets/)

[Business](https://www.cnbc.com/business/)

[Investing](https://www.cnbc.com/investing/)

[Tech](https://www.cnbc.com/technology/)

[Politics & Policy](https://www.cnbc.com/politics/)

[Video](https://www.cnbc.com/tv/)

[Watchlist](https://www.cnbc.com/watchlist/)

[Investing Club](https://www.cnbc.com/investingclub/subscribe?__source=investingclub|globalnav|join&tpcc=investingclub|globalnav|join)

Join IC

[PRO](https://www.cnbc.com/application/pro?__source=pro|globalnav|join&tpcc=pro|globalnav|join)

Join Pro

[Livestream](https://www.cnbc.com/live-tv/)

Menu

[Tech](https://www.cnbc.com/technology/)

# Anthropic and OpenAI hunt for smaller data center deals, sources tell CNBC, in race to deploy AI capacity

Published Fri, Sep 18 20265:38 AM EDT

thumbnail

[Kai Nicol-Schwarz](https://www.cnbc.com/kai-nicol-schwarz/)[@in/kains](https://linkedin.com/in/kains)

[WATCH LIVE](https://www.cnbc.com/live-tv/)

Key Points

- Anthropic and OpenAI are exploring opportunities for smaller data center deals, sources told CNBC.
- Both companies are racing to deploy AI capacity and have announced a flurry of AI infrastructure deals over the past year as demand booms.
- Smaller capacity deals are often attractive because of "speed to usable capacity," one analyst told CNBC.

In this article

- [NVDA](https://www.cnbc.com/quotes/NVDA)

Follow your favorite stocksCREATE FREE ACCOUNT

Anthropic and OpenAI are hunting for smaller AI data center deals, sources told CNBC, as the race to access the infrastructure needed to deploy workloads ramps up.

The two AI labs have both inked [huge deals](https://www.cnbc.com/2026/07/27/nvidia-and-openai-in-talks-for-up-to-250-billion-dollar-ai-backstop.html) for AI data centers in the past year for facilities of multi-hundred-megawatt and gigawattcapacity, but sources have said those companies are now also looking for compute capacity deals for much smaller deployments of 20-30 MW.

Anthropic has sounded out agreements within that range across the U.K. and the Nordics, four people familiar with the conversations, who asked to remain anonymous when discussing private business dealings, told CNBC. OpenAI had been exploring opportunities for those smaller capacity deployments in the Nordics, two of the sources said.

One source said they were also familiar with talks involving Anthropic and OpenAI about U.S. capacity deployments at that scale.

Both companies have announced a flurry of AI infrastructure deals over the past year as they've looked to train and serve their models to end users. Deals to secure smaller allocations of compute allow companies to deploy workloads faster amid the AI boom.

"We're building a diversified compute portfolio to meet growing demand for AI around the world," an OpenAI spokesperson told CNBC.

"Different workloads need different infrastructure, so we have conversations with a range of partners and assess opportunities based on our requirements, performance, reliability, timing and cost," they added. "We don't comment on specific commercial discussions."

Anthropic did not comment when approached by CNBC.

## 'Speed to usable capacity'

Both AI labs typically rent compute capacity from data center operators and neoclouds and have sought large-scale, long-term agreements.

Anthropic inked a roughly [$45 billion cloud deal](https://www.cnbc.com/2026/08/26/anthropic-and-nscale-strike-45-billion-cloud-deal-sources-say.html) with Nscale, which will see the AI lab rent around 460 MW of compute capacity at a data center development in West Virginia, two people familiar with the matter told CNBC in August.

OpenAI has said it surpassed the original commitment of 10 GW to its Stargate AI infrastructure project in April and has since committed to developing a further 3 GW in Georgia and [8 GW in Ohio.](https://www.cnbc.com/2026/08/17/nvidia-financing-open-ai-data-center-ohio.html)

Huge data center projects in the U.S. and further afield are [increasingly facing pushback](https://www.cnbc.com/2026/09/06/ai-data-centers-are-transforming-rural-land-markets-fueling-backlash.html) from local communities. The sector is also under pressure in much of Europe, where available land and power are in short supply.

CoreWeave CEO: AI industry has not done a good job explaining data center impact on communities

watch now

VIDEO4:2204:22

CoreWeave CEO: AI industry has not done a good job explaining data center impact on communities

[Squawk on the Street](https://www.cnbc.com/squawk-on-the-street/)

Smaller capacity deals are often attractive because of "speed to usable capacity," Jabez Tan, head of research at Structure Research, told CNBC.

"Securing a few megawatts at an existing powered site can be more practical than waiting for a much larger block in one location," he said. "For workloads that can operate across separate sites, a collection of smaller deployments can add up to substantial capacity."

## Shift to inference

Training AI models requires large amounts of computing power to process huge quantities of data, but deploying those systems day-to-day — a processknown as inference — can be done with smaller clusters of chips.

"Training a large model typically requires many chips working closely together," Tan said. "Many inference workloads can instead serve separate requests across multiple smaller clusters, opening up more locations."

The shift matters as more AI compute moves from training models to serving them in production. The amount of capacity being used to serve inference is therefore expected to rise.

The proportion of total data center capacity used for inference workloads is expected to overtake training workloads in 2027, according to a report by real estate company JLL. In 2025, inference made up 9% of global workloads in data centers compared to 14% for training, the report said. By 2030, inference is projected to use 37% of that capacity, compared to just 13% for training.

In February, it was announced that [Nvidia](https://www.cnbc.com/quotes/NVDA/) would collabora
cebert81
🟧 echo.other ⭐CNBC’s original report says Anthropic and OpenAI are seeking 20–30 MW data-center deals in the U.K., Nordics, and potentially the U.S., baseKai Nicol-Schwarz (CNBC)——

Interpretation history

Decision trace