AI Infrastructure news

OpenAI, Softbank, and Blackstone have formed a new lobbying group to push for more data center construction. The American Infrastructure Alliance will fight local opposition, aiming to secure the physical infrastructure needed for AI's continued growth.

AMD has released its most powerful and expensive server CPU ever, the EPYC 9996. It packs 256 cores, consumes 600W of power, and supports next-generation technologies like PCIe 6.0, targeting high-performance computing and AI workloads.

Google Cloud has launched a new, lower-cost storage tier for its Managed Lustre service. This makes high-performance file systems more affordable for AI and HPC workloads by reducing the need to manually manage different storage types.

Google Cloud's 'fluid compute' strategy helps teams avoid AI hardware shortages. It lets developers design workloads that can flexibly run on different available accelerators, like GPUs or TPUs, preventing costly project delays.

Intel researchers developed a new storage format that compresses AI models smaller than previously thought possible. This method boosts performance by up to 27% on GPUs without needing to retrain the model, making AI more efficient.

The UK's £100 billion data center boom is hitting a wall. A new survey finds that cabling bottlenecks and the rapid pace of AI mean new facilities are often outdated the moment they go online, threatening the country's tech goals.

Kioxia is developing new flash memory products that act like traditional RAM. This could help data centers dramatically expand memory capacity for AI workloads without the high cost of DRAM, easing a major infrastructure bottleneck.

Amazon Web Services is massively expanding its AI infrastructure, planning to add two million more Nvidia GPUs by 2028. This move signals the immense, ongoing demand for AI computing power among cloud customers.

AI agents often fail outside of controlled demos because they lack a proper support system. An 'agent harness' provides the necessary infrastructure and guardrails to make them reliable and trustworthy for real-world use.

Kioxia unveiled a new enterprise SSD that achieves a record 10 million input/output operations per second. Its extreme speed and endurance are engineered to eliminate data bottlenecks for the most demanding AI systems.

Oracle co-founder Larry Ellison lost $104 billion in two months as the company's stock fell 54%. The drop reflects market uncertainty over Oracle's high-stakes, capital-intensive strategy to compete in the AI infrastructure race.

A growing bipartisan movement is successfully blocking new data center construction. This local opposition creates a significant physical and regulatory risk for the future of cloud and AI infrastructure, potentially driving up costs and limiting scalability.

AI startup Mirendil is spending over $100 million on Google Cloud to build self-improving AI systems. The deal highlights the massive computing power now required for next-generation AI research aimed at accelerating scientific discovery.

Vercel has significantly increased the capacity of its Sandbox feature for Pro and Enterprise users. The update allows for 10,000 concurrent sandboxes and a tenfold increase in CPU allocation, enabling more complex, large-scale applications like AI tools.

AI workloads are demanding so much power that traditional data centers can't keep up. This is forcing a complete rethinking of infrastructure, from energy supply to cooling, impacting future cloud costs and availability for everyone.

GPU cloud provider Nscale is buying Anyscale, the company behind the open-source Ray framework. The acquisition tests whether a critical AI tool can stay cloud-neutral, a major concern for developers building on any platform.

Weka has launched a new storage system that packs a claimed exabyte of data into a single server rack. It combines ultra-dense SSDs with powerful software compression to dramatically increase storage capacity for AI and big data workloads.

A data center firm is spending $500M to convert 15 former crypto mining sites into AI cloud facilities. The deal highlights the intense competition for the massive power and infrastructure needed to fuel the AI boom.

Data centers and crypto miners in Texas are failing crucial grid stability tests. This threatens the state's power supply and could lead to stricter rules and higher costs for large-scale computing and AI operations.

Simple vector search is no longer enough for production AI. Companies are now building hybrid systems that combine it with ranking and personalization to deliver more relevant and useful results.

xAI has signed a multi-billion dollar deal to provide its competitor, Anthropic, with large-scale AI computing services. The agreement, worth about $1.25 billion per month until May 2029, signals a major shift where specialized AI compute is emerging as a standalone business, challenging traditional cloud providers.