Editor's Summary

This was a week about AI moving from labs into the world it has to actually work in — new markets, new professions, and the harder problem of getting enterprises to use any of it.

Apple had the busiest week of any vendor. China's Cyberspace Administration approved Apple Intelligence for the domestic market, contingent on swapping in Alibaba's Qwen model in place of Apple's own US-trained systems, since the latter can't legally operate there — a reminder that frontier AI is increasingly balkanized by geography rather than shipped as one global product. Days later, Apple pushed its overhauled Siri into the iOS 27 public beta, giving ordinary users their first hands-on look at the personal-context-aware assistant ahead of its fall launch, warts (and beta bugs) included. Meanwhile Anthropic and Blackstone launched Ode, a $1.5 billion venture that embeds Anthropic engineers directly inside enterprise clients — a bet that the next big AI business isn't better models but the unglamorous work of actually implementing them inside real companies. Google, by contrast, chose to hold back: Gemini 3.5 Pro's launch was delayed after the model missed internal goals, notably on coding performance, a rare public admission that "ship it anyway" isn't the strategy at every lab.

Three smaller stories rounded out the week. Mira Murati's Thinking Machines Lab released Inkling, its first open-weight model — a 975-billion-parameter mixture-of-experts system trained on 45 trillion tokens — explicitly betting that customizable, organization-specific models will beat one-size-fits-all frontier systems. Applied Computing raised a $20 million Series A for Orbital, a foundation model aimed squarely at oil, gas, and petrochemical plants, arguing operators currently use less than 8% of their sensor data and that AI can compress days-long investigations into seconds. And Anthropic launched Claude for Teachers, a free K-12 offering tied to standards-mapped curricula across all 50 states, extending its enterprise-implementation push into classrooms.

Taken together, the week's real story is less about capability jumps and more about distribution and trust: who gets to run these models where, who helps companies actually deploy them, and who's willing to slow down rather than ship something short of the bar. Worth watching next week is whether Gemini 3.5 Pro's delay stretches further, and how quickly Apple's Qwen-powered rollout in China actually reaches users.

Sunday, July 19, 2026
Can an Apple lawsuit derail OpenAI's hardware plans?
Nonprofit Current AI is racing to build the World Wide Web of AI, free for all
What to watch for after Jensen Huang's Japan visit
'Odyssey' director Christopher Nolan calls AI an obvious 'Trojan horse'
Thursday, July 16, 2026
Google Gemini Launch Delayed as Tech Falls Short of Internal Goals
Moonshot's upcoming Kimi 3 is expected to close the gap with Anthropic's Opus 4.8
Google's AI Mode now lets you link and interact with select apps
Roblox launches an AI-powered game-creation feature in its mobile app
Why AMI Labs' Alexandre LeBrun won't call his AI 'AGI' or 'superintelligence'
Google Vids now lets you star in your own AI videos
Wednesday, July 15, 2026
Apple Intelligence approved for launch in China with Alibaba's Qwen AI
Anthropic and Blackstone launch Ode, a $1.5 billion AI implementation venture
Thinking Machines unveils Inkling, its first open-weight model, betting against one-size-fits-all AI
Applied Computing wants to give oil and gas operators an AI model for the entire plant
Amid hardware legal battle, OpenAI releases a $230 keyboard for Codex
Indian AI coding startup Emergent becomes a unicorn just over a year after launch
Tuesday, July 14, 2026
Apple opens its new Siri AI to everyone with the iOS 27 public beta
Claude for Teachers