Full archive

Infrastructure AI News

Search reporting on AI markets, models, products, companies, policy, and culture.

Showing 1-18 of 148

Today

14 stories

36 minutes ago

InferenceX Puts GB300 First for MiniMax M2.7, but Only on a Fixed Chat Test

The new ranking gives infrastructure teams a comparable speed and cost snapshot, but its single-turn design does not test the multi-step agent work MiniMax highlights for M2.7. MiniMax pitches M2.7 for complex agent harnesses and software-engineering work. GB200 NVL72 came second at 12,515 tokens per second per GPU, a 10% gap.

2 min readRead story

Infrastructure

2 hours ago

South Korea’s Chip Exports Hit $46.65 Billion as AI Demand Raises Concentration Risk

August’s record semiconductor shipments show how strongly cloud-provider AI investment is reaching the supply chain. They also put more weight on whether demand cools gradually or suddenly while monetary policy remains tight. South Korea’s semiconductor exports rose 209% from a year earlier to a record $46.65 billion in August, turning AI infrastructure spending into an unusually concentrated export windfall.

3 min readRead story

Platform shift Infrastructure

3 hours ago

Broadcom Forecasts $230 Billion in AI Chip Revenue for Fiscal 2028

The long-range target would put Broadcom on a far larger AI-chip trajectory. Its fourth-quarter guidance is the next concrete measure of whether that momentum continues. Broadcom CEO Hock Tan forecast that the company’s AI-chip revenue will reach about $115 billion in fiscal 2027 and $230 billion in fiscal 2028.

2 min readRead story

Enterprise adoption Infrastructure

4 hours ago

Enterprise Buyers Put Non-Nvidia Chip Evaluations Ahead of Nvidia’s Next GPUs

The July survey signals a broader search across AWS, Google, AMD, Intel and custom hardware. It does not show a migration away from Nvidia, but it does show buyers weighing performance, reliability and integration at the workload level. Enterprise AI buyers are widening their chip searches without signaling a fast break from Nvidia.

3 min readRead story

Partnership Infrastructure

6 hours ago

Google Agrees to Buy 396 MW of Geothermal Power for a Planned Utah Data Center

The agreement gives Fervo a defined customer for Cape Station while tying Google’s future data-center supply to a geothermal project still under construction. Google has agreed to buy 396 megawatts of electricity from Fervo Energy’s Cape Station geothermal plant in Utah for a planned data center. Google also has an option to purchase another 600 MW through June 2030, from a project expected to come online in 2028.

2 min readRead story

Government action Infrastructure

11 hours ago

Nvidia’s Huang Urges G20 Countries to Build AI Infrastructure for Economic Growth

The Nvidia chief’s pitch puts domestic data centers at the center of national AI strategy. Nvidia’s Ohio project illustrates the physical requirements behind that argument: land, electricity and buildings alongside computing hardware. Jensen Huang wants the world’s largest economies to treat AI capacity less like a software purchase and more like a utility buildout.

2 min readRead story

Launch Infrastructure

12 hours ago

Equinix Sets Late-2026 Fabric One Beta for AI and Multicloud Links

The planned service would put a managed orchestration layer over connections among enterprise systems, clouds and AI environments. The key test is whether it can turn broad customer requirements into dependable routing, security and recovery controls. Equinix is planning to make a hard-to-assemble layer of AI infrastructure a managed service.

2 min readRead story

Launch Infrastructure

13 hours ago

Databricks Pushes Lakebase Postgres as the Backend for Apps and Agents

The developer push pairs managed PostgreSQL with Databricks Apps, handling application identities and connection details inside the workspace while leaving teams to choose their database environment and data-access model. Databricks is positioning Lakebase Managed Postgres as a managed PostgreSQL backend for web applications and AI agents. Connection details are injected into the application as environment variables.

3 min readRead story

Partnership Infrastructure

13 hours ago

Equinix Sets Q1 2027 Start for Distributed Enterprise AI Inference Service

The planned service combines NVIDIA infrastructure designs, Together AI model serving and Equinix connectivity. Its test is whether one package can simplify placing AI workloads across providers and locations. Equinix has announced Inference Exchange, a planned service for enterprises running AI workloads across locations, clouds and providers.

2 min readRead story

Infrastructure

14 hours ago

Berkshire’s Greg Abel Says Hyperscalers Must Cover Their Own Grid Costs

The condition puts cost allocation at the center of Berkshire Hathaway Energy’s AI-infrastructure strategy as local opposition increasingly focuses on electricity bills and resource use. Berkshire Hathaway CEO Greg Abel says the company will serve hyperscale data centers only if their power needs do not raise rates for Berkshire Hathaway Energy’s existing customers.

3 min readRead story

Partnership Infrastructure

15 hours ago

SpaceX’s External Compute Deals Add a Potential Rival for CoreWeave

A confirmed Anthropic arrangement and reported outside-customer contracts put more large-scale AI capacity into the market. The unanswered question is whether it can become a repeatable cloud service. CoreWeave may face a new kind of potential AI-cloud rival: SpaceX -linked computing capacity that is beginning to reach external customers, not only xAI’s own workloads.

2 min readRead story

Infrastructure

15 hours ago

Loudoun County Confronts Data-Center Backlash After Reaping a Tax Windfall

Virginia’s Data Center Alley shows the bargain confronting new AI-infrastructure hosts: commercial revenue can transform public services, while the physical costs become harder for residents to ignore. Loudoun County, Virginia, built one of the world’s largest concentrations of data centers into a local fiscal engine. The result is a strong case for hosts seeking an expanded commercial tax base.

3 min readRead story

Infrastructure

19 hours ago

Musk Warns of AI Power Shortfall as Zuckerberg Cites Data-Center Labor Gap

At the G20 Innovation Ministerial in Chapel Hill, two leading AI infrastructure backers pointed to limits outside the data center: available electricity and the people qualified to build facilities. Elon Musk warned G20 technology ministers that AI could face a significant power shortfall next year. The G20 Innovation Ministerial runs September 1–2 at the Carolina Inn in Chapel Hill.

3 min readRead story

Infrastructure

20 hours ago

U.S. Data-Center Construction Hits $75B Annual Pace as AI Buildout Meets Bottlenecks

July’s construction surge covers buildings and integrated site equipment, not the far costlier servers and power systems still needed to put new AI capacity online. U.S. data-center construction spending reached a seasonally adjusted annual rate of $75 billion in July, after rising 6.2% from June and 57% from a year earlier. That distinction is central to the stakes.

3 min readRead story

Yesterday

4 stories

Research Infrastructure

Yesterday

Baseten’s New Inference Guide Separates Configuration Tradeoffs From True Efficiency Gains

The guide’s useful test for infrastructure teams is whether a change merely reallocates latency, throughput and cost—or reduces the work required to serve every token. Not every faster-looking LLM deployment is a true efficiency gain. The distinction separates a workload decision from a reduction in the resources needed for serving.

3 min readRead story

Government action Infrastructure

Yesterday

Army’s $10B ITES-4H Tests Whether Faster AI Infrastructure Buying Can Keep Its Guardrails

A new expert debate frames the Army’s existing hardware vehicle as more than a purchasing channel: it is a test of whether shared standards can speed AI readiness without reducing competition and technical diligence to formalities. The Army’s $10 billion ITES-4H contract is meant to provide a streamlined route to the servers, storage and networking behind AI workloads.

2 min readRead story

Earnings Infrastructure

Yesterday

Dell Lifts AI Server Goal to $74B; Shares Rise 10% After Earnings and Outlook

The higher target puts a concrete annual delivery marker on Dell’s AI infrastructure business. Its reported quarter establishes the starting point; future server sales must now substantiate the forecast. Dell lifted its full-year AI-optimized server sales target to $74 billion after fiscal second-quarter sales reached $16.40 billion.

3 min readRead story

Acquisition Infrastructure

Yesterday

GoPro’s $285M Starman Merger Targets AI Data Centers, Keeps Cameras

Starman Optical’s transceivers are GoPro’s proposed bridge into data-center infrastructure, while shareholder payouts, debt repayment and the new ownership structure remain contingent on closing. GoPro is proposing an AI-infrastructure expansion without giving up its camera business. The deal connects a specific photonics business to GoPro’s data-center ambition.

2 min readRead story

Load more stories18 more next