Developing storyline

AI infrastructure buildout: chips, clusters, and mega-deals

Tech giants, chipmakers and crypto-miners race to expand AI computing capacity through new chip designs, data-center leases, and capital raises as demand for AI infrastructure surges.

6 stories516 sourcesUpdated Aug 11, 2026

In this storyline:Nvidia Taiwan Semiconductor Manufacturing Co.Jensen Huang AMD Ventures anthropic principle Intel OpenAI Red Hat OpenShift

Coverage momentum

total articles across this storyline · last 72h

Key developments

Aug 11, 2026
Nvidia and banking partners are mobilizing roughly $500 billion for AI infrastructure as Broadcom emerges as a credible rival with custom accelerators.
220 sources on this →
Aug 11, 2026
IBM and Nvidia expand AI infrastructure on IBM Cloud, while IBM and Together AI ink a $240 million deal for a large-scale inference cluster.
22 sources on this →
Aug 11, 2026
Nvidia unveils the open-source Nemotron 3.5 Lightning model and open-sources cuFile APIs to broaden access to its AI stack.
34 sources on this →
Aug 11, 2026
Riot Platforms signs a 20-year, 191 MW data-center lease with a frontier AI lab at Rockdale, potentially worth up to $9.1 billion.

Coverage across the spectrum

Claim check

What each side asserts, disputes — or leaves out entirely.

Analyzing the coverage…

Stories in this storyline

Photo: ZMASLO / Wikimedia Commons (CC BY 3.0)

Nvidia, Banks to Mobilize $500B AI Infrastructure

Nvidia remains the dominant AI platform with a powerful data-center and networking moat, while Broadcom is emerging as a credible challenger through custom AI accelerators and AI networking, supported by strong AI-revenue trajectories and high-profile deals like OpenAI. Investors are weighing concerns about financing-driven demand versus durable growth as Nvidia pursues strategic bets such as Safe Superintelligence and expansion into AI base stations, potentially widening opportunities for domestic suppliers.

220 sources

NVIDIA Expands Nemotron, Introduces Switchyard Router

Nvidia unveiled Nemotron 3.5 Lightning, a lightweight open-source 30B MoE model that can run on a single GPU and is designed for fast, customizable local AI agents and task-specific use cases, with open weights enabling fine-tuning. The company also open-sourced cuFile APIs and the supporting storage stack to streamline direct GPU access in AI pipelines, reflecting its broader open AI strategy and push toward AI factories as CEO Jensen Huang defends open models.

Comments

Ask this storyline anything:

Free account · your comment posts right after signup

73 sources on this →

Aug 11, 2026

Microsoft accelerates development of its Maia 300 AI chip with TSMC, aiming to reduce reliance on Nvidia by 2027.

33 sources on this →

Aug 11, 2026

Intel plans a stock offering of up to $20 billion to fund AI chipmaking and advanced manufacturing expansion.

157 sources on this →

34 sources

IBM-Together AI to Build Inference Cluster

IBM and NVIDIA are expanding AI infrastructure on IBM Cloud with HGX B300 systems powered by Blackwell Ultra GPUs, integrated with Red Hat OpenShift and watsonx for governed hybrid-cloud workflows, and IBM targets serverless GPU access in 2026. In a parallel push, IBM and Together AI announced a $240 million multi-year agreement to deploy a large-scale inference cluster on IBM Cloud using HGX B300 and Spectrum-X networking, aiming for open-source model inference with substantially higher throughput and availability in early 2027.

22 sources

Photo: Coolcaesar / Wikimedia Commons (CC BY-SA 4.0)

Intel Raises Stock Offering to Up to $20B

Intel plans a roughly $15 billion underwritten public stock offering to fund AI and advanced manufacturing expansion, with a 30-day option for underwriters to purchase up to an additional $2.25 billion in shares. The move highlights the capital-intensive push in AI chipmaking, ongoing turnaround efforts, and potential dilution concerns for existing shareholders amid government and industry investments in domestic semiconductor capacity.

157 sources

Riot Inks 20-Year, 191 MW Anthropic Rockdale Lease

Riot Platforms reported Q2 2026 revenue of $174.2 million and completed the initial 25 megawatts of AMD deployment, while signing a 20-year lease for 191 MW of data-center capacity at Rockdale with a frontier AI lab, bringing contracted capacity to 241 MW and potential revenue up to $16.1 billion with extensions. The 191 MW lease is expected to generate about $9.1 billion over the initial term, with full deployment planned by mid-2028, supported by interim financing and ongoing expansion alongside the AMD agreement and bitcoin mining operations.

73 sources

Photo: See the ticket in the Permission field below. / Wikimedia Commons (CC BY-SA 3.0)

Microsoft Targets Maia 300, Over 1 Million Units

Microsoft is accelerating its Maia 300 AI chip, aiming to mass-produce with TSMC for delivery in 2027 and to persuade cloud customers like Anthropic to adopt it, signaling a push to reduce dependence on Nvidia. Despite earlier delays, Maia 200 is already deployed in data centers and Maia 300 remains in design with a potential public unveiling soon as Microsoft expands its in-house accelerator lineup.

33 sources