Microsoft Patent Proposes Rewards for AI Training on Idle GPUs
Microsoft’s patent describes opt-in AI training on spare GPU capacity, with possible credits redeemable for software, model access or money.
Access a continuous stream of concise, high-impact reporting on artificial intelligence across global markets and industries. This section aggregates short-form coverage of corporate announcements, funding activity, product launches, regulatory actions, executive moves, and strategic partnerships, delivering essential developments with speed and clarity. Designed for efficient consumption, it prioritizes material updates and market-moving information while complementing longer-form analysis and in-depth industry coverage.
Microsoft’s patent describes opt-in AI training on spare GPU capacity, with possible credits redeemable for software, model access or money.
David Solomon expects strong AI infrastructure financing over the next 12–24 months as computing demand exceeds supply, while warning of setbacks.
Foxconn chairman Young Liu sees potential for 50% AI server market share within a couple of years, with capacity booked for at least 18 months.
Anthropic’s S-1 details up to $120.5 billion in SpaceX.AI compute spending and a minimum $114.5 billion Google commitment through July 2033.
Elon Musk predicts failures and consolidation among AI cloud providers as debt financing tightens, while Nebius secures a $3.5 billion Citi facility.
DeepSeek is reportedly working with Huawei on chips tailored to its AI models, alongside an expanding software ecosystem for Ascend processors.
Citi says AI returns depend on infrastructure and enterprise deployment catching up with models, as data center costs rise and many firms remain in pilots.
Volantis is developing photonic links between processors and memory, with its first customer systems planned for 2027.
Japan's planned AI infrastructure expansion involves Dell and JERA, with a first project in Chiba targeting up to 400 megawatts of capacity.
Nvidia has authorized an additional $150 billion in share repurchases, bringing its total buyback capacity to $235 billion, the largest single authorization increase in U.S. history.
Armenia has emerged as a regional AI infrastructure hub following the opening of Firebird's Nvidia-powered AI factory in Hrazdan, the first of its kind in the CIS.
Palantir Technologies has named Nebius Group its preferred sovereign AI infrastructure partner, with plans to integrate Nebius compute and inference endpoints directly into the Palantir enterprise perimeter.