NVIDIA's new revenue-share model pulls in Sharon AI's 40,000 GB300s and a 170,000-GPU Firmus campus in Batam. The chipmaker now wants recurring cloud rent, not just silicon margin.
Workday's Agent Passport ships Q3 2026 in early access with exactly one testing partner, Cisco, and no answer yet on who pays when a vetted agent misbehaves.
Base44 just rolled out Base1, its in-house LLM trained on user data, betting that vertical integration beats renting Claude Opus by the token.
Two of Google DeepMind's biggest names walked out in 48 hours. The market wiped 5% off Google's shares. The engineering question is what that exodus actually signals.
OpenAI and Broadcom unveiled Jalapeño, a blank-slate LLM inference ASIC taped out in nine months and aimed at gigawatt-scale deployment with Microsoft by end of 2026.
Anthropic dropped Claude Tag into Slack this week, replacing the old chatbot with a multiplayer agent powered by Opus 4.8. Here's what it breaks and what it fixes.
Sakana AI shipped Fugu, an orchestration model that routes tasks across a swappable pool of frontier LLMs. Early reaction skews skeptical. Here's what holds up.