OpenAI co-founder Greg Brockman's personal notes reveal that they've been discussing turning the AI firm into a non-profit without Elon Musk.

Leading developers of AI models from China want Nvidia's Rubin and explore ways to rent the upcoming GPUs in the cloud.

A new report says Nvidia's presence is about to shrink drastically in China.

One Japanese manufacturer has found itself at the center of a new AI-driven supply shortage. Glass cloth is a crucial component for many components, and with dwindling supplies, companies are looking for alternatives.

U.S. introduces new export rules for AI and HPC accelerators developed in America that allow shipments of AMD's MI325X and Nvidia's H200 GPUs to Chinese entities, but only if their developers meet demand of American customers and meet other requirements of the Department of Commerce.

KIs sollen die Arbeit in Forschungslaboren revolutionieren, doch eine aktuelle Studie warnt vor tödlichen Risiken. Ein neuer Benchmark offenbart, dass selbst moderne Top-Modelle bei grundlegenden Sicherheitsfragen versagen und gefährliche Anweisungen geben. (Weiter lesen)

OpenAI greift Google und DeepL an: Mit ChatGPT Translate startet ein neuer Web-Dienst für Übersetzungen. Er punktet nicht mit der Masse an Sprachen, sondern mit Kontextverständnis und Tonfall-Anpassungen. Kann der KI-Pionier die Platzhirsche verdrängen? (Weiter lesen)

Amazon has struck a two-year deal to receive copper from an Arizona mine, for use in its AWS data centers in the U.S.

The U.S. may allow shipments of rather powerful AI processors to China on a case-by-case basis, but with the U.S. supply priority, do not expect AMD or Nvidia ship a ton of AI GPUs to the People's Republic.

Phison's aiDAPTIV+ stack enables large Mixture of Experts AI models and agentic AI workloads to run on client systems with limited memory capacity.

After weeks of deliberation, sources suggest that China will only allow companies to purchase H200 GPUs for "special circumstances," although Beijing has yet to define exactly what that means.

China’s AI sector made a splash this month as a cluster of domestic AI firms raised more than $1 billion through IPOs in Hong Kong.

A new Deepseek whitepaper has outlined a new form of long-term memory for AI models, named Engram. Engram-based models are more performant than their MoE counterparts, and decouple compute power from system RAM pools to improve results.

Microsoft's five-point plan focuses on electricity use, water consumption, building local jobs and training people, investing in local infrastructure, and the long-term skills development of the community surrounding its AI data centers.

The U.S. President says that his team will work with various AI tech companies, starting with Microsoft, to ensure that their power demands will stop affecting the power prices that ordinary Americans pay for.

The U.S. House of Representatives has passed a bipartisan bill, expanding U.S. export controls to stop China from using offshore data centers to access banned chips.

Meta establishes 'Meta Compute' organization to manage buildout of next-generation data centers and development of custom computing architectures that will power gigawatt-class data centers.

The UK is making it a crime to generate or request AI-made explicit content from this week, following the ban on sharing deepfakes. The region's communications regulator, Ofcom, is also looking into Grok, investigating the service formally to see if it "has complied with its duties to protect people."

Nvidia's Jensen Huang says negative narratives around AI are "extremely hurtful," and that science fiction speculation isn't connected to reality.

The 6GW will come from various suppliers — both existing and upcoming — to deliver the estimated 1GW that Meta's Prometheus site will require in 2026, and the 5GW its Hyperion data center will need by 2028.