Strategy
118 analyses
Google launched Gemini 3.6 Flash at $1.50 per million input tokens and $7.50 for output, generating 17% fewer tokens. On the same day, DeepMind confirmed the start of the largest pre-training in history.
Three days after launching the 2.8 trillion parameter model, Chinese company Moonshot suspended new users due to GPU limits. Current subscribers remain unaffected as the company moves toward a listing in Hong Kong.
The preview of the 2.4 trillion parameter model arrives three days after Moonshot's Kimi K3. Alibaba's stock surged by up to 5.4% in Hong Kong, re-establishing China at the forefront of AI discussions.
At the WAIC in Shanghai, Alibaba responded to Moonshot's Kimi K3 with a preview sold at 10% of the standard price. No model card, no active parameters disclosed.
With banking clients collectively cutting over 10,000 jobs in Q2 2026, the world's four largest auditing firms are deploying AI agents in work that has trained generations of accountants.
For nearly 48 hours, an error in the pricing subsystem of Amazon Web Services displayed fictitious billion-dollar charges; Amazon completed the fix on July 18 at noon Pacific Time.
With 2.8 trillion parameters and a context of 1 million tokens, Kimi K3 has achieved first place in the Frontend Code Arena benchmark, surpassing Claude Fable 5. Weights will be released on July 27 for local deployments.
4 billion parameter model runs on Jetson at the edge. Coalition brings together AIRoA, FANUC, Fujitsu, Hitachi, Kawasaki, Kubota, NEC, SoftBank, Sony, and Yaskawa around robotics.
Salesforce, Workday, Adobe, and ServiceNow also fell in the wake of IBM's warning. The easy thesis is an AI bubble. The correct reading is a budget reorganization that no one rehearsed.
The new Ode with Anthropic brings together $1.5 billion from Anthropic, Blackstone, Hellman & Friedman, and Goldman to sell AI implementation to the same clients as the Big 4.
The second extension in a week comes after the previous deadline has passed and on the day Sol begins operating for OpenAI's paid subscribers. It's an urgency discount disguised as courtesy.
The Information details on Monday the coalition promoting ARD as an alternative to MCP. The unspoken gain is to reduce the implicit tax of running agents on a competitor-standardized stack.