It's Monday, August 17th: Apple trained its own AI model for China and filed the service with Beijing's regulator, and SpaceX closed its $60 billion purchase of the code editor most of the Fortune 500 runs.
Covering what’s happening on the ground in AI, every Monday.
1️⃣ CHINA FILING: Apple Trained Its Own Model, And Beijing Registered The Service
Image from CNA
Reuters reports Apple trained its own large language model for the China market with Alibaba's help, and that China's internet regulator registered Apple's generative AI service in July. Reuters says that if the model ships, Apple would be the first foreign company Beijing has cleared to run its own model on the mainland.
Reuters reported on August 14, citing three people familiar with the matter, that Apple built the model itself with technical support from Alibaba, which Apple picked to power its China phones back in February 2025, when chairman Joe Tsai revealed the deal in Dubai.
China's Cyberspace Administration registered Apple's generative AI service in July, in a batch of seven on-device filings that also covered Huawei, Xiaomi, OPPO, vivo, Samsung and Nubia.
Apple Intelligence is expected in China within months through an iOS update, expected to run Apple's model alongside Alibaba's Qwen and some Baidu components. Apple posted a Chinese support page on August 8 showing Mac owners how to route Siri and Writing Tools through Qwen, then pulled it inside a day, MacRumors reported.
Apple took 18.1% of China smartphone shipments last quarter, up from 13.9% a year earlier, according to IDC, while Huawei held the lead at 22.6%.
Apple's route is the one other foreign teams will now copy, filing through a Chinese legal entity for the service rather than the weights, and keeping a domestic model wired in alongside anything they trained themselves. Watch the iOS update that switches Apple Intelligence on there, because the feature set Apple ships in China will show what a foreign-trained model is actually permitted to do.
2️⃣ SPACEX CLOSES CURSOR: A $60 Billion All-Stock Deal Moves The Code Editor Inside A Rocket Company
Image from Cursor
SpaceX completed its purchase of Anysphere on August 14, converting Cursor stock into 389,289,254 SpaceX Class A shares and making the code editor a wholly owned subsidiary of a publicly traded launch company.
The merger became effective August 14, two months after the June 16 agreement was signed. Cursor's implied equity value was fixed at $60.0 billion, and the share count came from dividing that by the volume-weighted average closing price of SpaceX stock over the seven trading days before close. SpaceX assumed roughly 44.4 million Cursor stock options on top of the shares.
Cursor says the product continues and now runs on what it calls the largest fleet of GPUs in the world, which it expects to turn into more capable models at lower cost.
SpaceX told shareholders in its second-quarter results that the deal accelerates its AI enterprise offerings, and that Grok 4.5 was trained alongside Cursor. That AI segment booked $2.6 billion in quarterly revenue, up 247% year over year, none of it from Cursor, which had not yet closed.
This deal puts the model vendor and the editor under one roof. Cursor lands inside SpaceXAI, the division formed after SpaceX absorbed xAI in February, so Grok becomes an in-house model for an editor that Michael Truell co-founded and that Cursor says 64% of the Fortune 500 now uses.
If Cursor sits in your daily loop, the roadmap for your editor now answers to a Nasdaq-listed company whose core businesses are launch and satellite internet, and which builds in-house the models the editor will increasingly default to. Watch the model picker and your invoice over the next two release cycles, and keep a second editor configured so a pricing or default-model change does not land on you mid-sprint.
📰 Other Headlines
HALF-PRICE FLASH: Google shipped Gemini 3.7 Flash three weeks after 3.6 Flash, lifting DeepSWE v1.1 from 49.0% to 65.3% at an introductory $0.75 per million input tokens.
DEEPSEEK REPRICES: DeepSeek moved its V4 family to peak and off-peak pricing, pushing V4-Flash output from a flat $0.28 per million tokens to $1.32 at peak.
WATERMARKS INCOMING: Anthropic said future Claude models will embed a statistical watermark in generated text to comply with the EU AI Act, with a detection API still being finalized.
750 TOKENS A SECOND: OpenAI opened a limited preview of an Ultrafast tier running GPT-5.6 Sol on Cerebras hardware at up to 750 output tokens per second, roughly 14 times standard speed.
IBM GOES ELITE: IBM joined OpenAI's Elite partner tier and will embed GPT-5.6 and Codex into IBM Consulting Advantage, with thousands of consultants being certified.
GROK 4.6 SHIPS: SpaceX's AI division released Grok 4.6 at $2 per million input tokens and $6 per million output, scoring 61 on the Artificial Analysis Intelligence Index and matching GPT-5.6 Sol on that composite.
QWEN OPENS UP: Alibaba's Qwen team published open weights for Qwen3.8 under Apache 2.0, with the 27B model handling images and video across a 262,144-token native context.
🫵 Want your message in front of 200,000 AI builders?
Our partners and sponsors get exclusive placements across the newsletter and access to AIC's in-person network — demo nights, dinners, hackathons, and forums across 180+ chapters.
For all inquiries, send us a note at [email protected].
The AI Collective is built by volunteers across 180+ chapters in 40 countries.
Thank you to the thousands of volunteers around the world who make this work possible. We truly could not do this without you.
🧑💻 About the Editors
About Noah Frank
Noah is a researcher, innovation strategist, and ex-founder thinking and writing about the future of AI and the workforce. His work and body of research explores the economics of emerging technology and organizational strategy. Outside of AIC, Noah heads research for Centaurian AI.
About Lindsay Gross
Lindsay is an AI engineer, researcher, and writer focused on how AI systems behave in practice and what it takes to make them safe. Her work sits at the intersection of AI safety, governance, and product design, and at AIC she writes about the questions that matter most as these systems scale.