Share
Linkedin
Twitter
Facebook
Whatsapp
Copy Link
Amazon Web Services is in talks to sell its Trainium AI chips to other companies for use in data centers, marking a major shift in strategy. The move could create a $50 billion business that directly competes with Nvidia's AI chip dominance. AWS AI chief Peter DeSantis confirmed early-stage discussions are underway, driven by surging demand that has seen Trainium3 capacity sell out almost instantly.
Space companies backed by Elon Musk's SpaceX and Jeff Bezos' Blue Origin are in preliminary talks with insurers about coverage for orbital AI data centers. The discussions mark early progress for an experimental industry aiming to bypass Earth's power constraints, though insurers face challenges modeling risks for this novel space-based infrastructure.
Palo Alto startup Architect Labs emerged from stealth with $24 million in seed funding to transform custom chip design using AI. The company aims to reduce the two-year timeline and hundreds of millions in costs currently required to design custom chips, directly challenging industry giants Broadcom and Marvell who generate tens of billions annually from this business.
ByteDance is spending more than $1 billion annually on Microsoft's AI and cloud services, primarily purchasing OpenAI models through Azure. The TikTok owner leads a wave of Chinese tech giants accessing American AI technology through Microsoft, even as OpenAI and Anthropic refuse direct sales to China. Azure's AI revenue in the region has tripled, raising questions about geopolitical implications.
China's securities regulator announced reforms to accelerate public listings for AI large-model companies and startups in quantum technology, nuclear fusion, and brain-computer interfaces. The move comes as major US AI firms prepare Wall Street IPOs, signaling Beijing's push for tech self-sufficiency amid growing competition with Washington.
SK Hynix has shipped samples of its latest HBM4E AI memory chips to major customers including Nvidia. The 12-layer chips deliver speeds of 16 gigabits per second per pin and offer over 20% better power efficiency than previous models. The move positions SK Hynix ahead in the race against Samsung and Micron to supply next-generation AI infrastructure.
Outgoing Apple CEO Tim Cook has warned that price increases across Apple's product lineup are unavoidable due to AI's insatiable appetite for memory chips. The global shortage of memory chips has caused component costs to surge fourfold, forcing even Apple—one of the world's most powerful buyers—to pass costs to consumers. The iPhone 17 Pro could see increases of $270 or more.
Amazon's AI chief Peter DeSantis openly acknowledged that the company's models haven't reached frontier level compared to OpenAI and Anthropic. Despite investing $33 billion in Anthropic, Amazon is building its own rival models through Nova2 and plans to compete with leading AI labs within the coming year using custom silicon and proprietary data.
Odyssey AI secured $310 million in Series B funding at a $1.45 billion valuation to advance AI world models that simulate the physical world. The round was led by Natural Capital with Amazon, AMD Ventures, and Google's GV participating. Notably absent was Nvidia, which backed the startup just four months earlier, as Odyssey partners with AWS and its Trainium chips instead.
The footwear company once valued at $4 billion has officially become Smartbird, an AI infrastructure firm. New CEO Nadia Carlsten, a former AWS executive, starts with $100 million in funding but zero employees, no customers, and no data centers. She's now tasked with building an AI compute business from scratch while investors watch closely.
Silicon Motion is aligning its consumer PCIe 6.0 SSD controller roadmap with Nvidia's bandwidth-hungry AI platforms rather than traditional CPU makers. The company plans to launch its Neptune controller in late 2027, but warns that AI data centers could consume up to 80% of global NAND production by 2027, creating supply constraints and higher prices for consumer storage.
AI chip startup SiMa.ai unveiled Palette Neat, the industry's first agentic development environment for Physical AI that reduces deployment timelines from months to days. The open-source tool uses natural language commands to eliminate low-level coding complexity, preserving 90% of legacy software investments while offering a pin-compatible alternative to Nvidia's edge AI solutions.
NVIDIA Blackwell swept every category in MLPerf Training 6.0, the industry's leading AI training benchmark. The platform achieved the fastest training times across all seven benchmarks, scaled to 8,192 GPUs, and was the only vendor to submit results for every test. GB300 systems delivered up to 60% faster performance than GB200, while competitors failed to submit results for newer workloads.
Israel's government has approved a comprehensive National AI Plan aimed at positioning the country as a global leader in artificial intelligence. Prime Minister Benjamin Netanyahu announced the initiative will integrate AI across all sectors while building sovereign computing infrastructure. The plan addresses Israel's AI catch-up revolution, with estimates suggesting up to four million workers may need to adapt their skills as the transformative technology reshapes the economy.
The Trump administration has postponed adding Chinese AI startup DeepSeek and more than 100 other firms to the Commerce Department's Entity List despite approval from an interagency committee. The decision aims to prevent further escalation of US-China tensions, even as concerns mount over DeepSeek's alleged support for China's military and intelligence operations and illicit extraction of capabilities from American AI models.
Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.