Intel has formally launched its GPU D270, an AI inference-focused graphics card offering up to 480GB of LPDDR5X memory, as the company works to establish a stronger position in the competitive AI accelerator market dominated by Nvidia and AMD. Built for Inference, Not Training Like Microsoft’s recently unveiled Maia 300, the D270 is positioned primarily…
Japanese automation company Omron has agreed to acquire OnLogic, a maker of rugged industrial computers, in a deal valued at $675 million, with the companies framing the combination as a move into “physical AI,” the application of AI systems to real-world industrial and robotic equipment. What OnLogic Brings to the Deal OnLogic specializes in industrial-grade…
NTT Data is in final-stage talks on a roughly $6.3 billion data center deal in Osaka, Japan, designed to supply 500 megawatts of capacity to support Anthropic’s growing compute needs, with both companies reportedly targeting a signing date of March 31. Scale of the Facility A 500 megawatt data center ranks among the larger single-site…
Nvidia’s share of China’s AI chip market has fallen to 34%, according to new data from research firm IDC, as domestic rival Huawei continues to close the gap with a 28% market share built on homegrown alternatives. The Numbers Behind the Shift IDC’s data shows Nvidia’s AI chip shipments into China fell 44% year-over-year, a…
Anthropic’s internal red team has warned that small open-weight AI models may be only months away from being capable of self-replication on consumer-grade internet-of-things hardware, a capability that has long been treated as a key warning sign in AI safety research. What Self-Replication Means in This Context Self-replication refers to an AI system’s ability to…
Microsoft has unveiled Maia 300, its third-generation custom AI accelerator chip, built specifically for inference workloads and equipped with 288GB of next-generation HBM4 memory, with broader deployment planned to scale through 2027. What Maia 300 Is Built For Unlike chips optimized primarily for training large AI models, Maia 300 is designed around inference, the process…
Meta has released two new generative AI models, Muse Wonder and Muse Wonder Full, expanding its push into AI-generated image and video content as competition in the space intensifies among major technology companies. What Muse Wonder Does The Muse Wonder family is built for generating and editing visual content, positioning Meta alongside OpenAI, Google, and…
Character.AI, one of the best-known consumer platforms for chatting with AI-generated personas, has filed for Chapter 11 bankruptcy protection, with Disney named as the stalking-horse bidder in a $20 million deal for the company’s intellectual property and transition services. The Filing Character.AI’s bankruptcy filing puts the company under court supervision as it restructures or sells…
The International Monetary Fund has formally flagged the financing structures behind the global AI buildout as a risk to financial stability, estimating that a sharp correction in AI-related investment could shave as much as 0.4% off global GDP by 2028. What the IMF Found In its latest assessment, the IMF pointed to the scale and…
Broadcom CEO Hock Tan told an audience in Singapore that the current wave of AI infrastructure spending “feels like a casino,” a remark that spread quickly through financial media and was followed within a day by a sharp selloff in Asian semiconductor stocks. What Tan Said Speaking at a technology conference, Tan reportedly described the…
