Good morning. It's Monday, July 20, and we're covering Japan's massive AI infrastructure push, China's new crackdown on AI companion apps, and why chip stocks rebounded after a bruising AI sell-off.

Plus: today's prompt on uncovering your "invisible work."

Apple broadens its lawsuit by warning dozens of former employees now at OpenAI to preserve documents and meet with its lawyers. Around 40 ex-Apple staff received legal letters as Apple alleges OpenAI misappropriated confidential hardware plans for its AI device efforts. The dispute threatens OpenAI's hardware ambitions ahead of its planned IPO and follows Apple's lawsuit naming two former employees.

Investors buy the dip after a sharp semiconductor selloff sparked by China's latest AI breakthrough. NVIDIA trims losses to 1%, while AMD, Broadcom, Intel, Marvell, Micron, and Sandisk rebound into positive territory. Global chip stocks have erased $3.3 trillion in value since June 22 as Moonshot AI's open-weight Kimi K3 and rising AI infrastructure costs weigh on sentiment. Earnings and hyperscaler AI spending now remain the sector's key focus.

Anthropic may become a customer rather than a competitor as it seeks scarce AI computing power from Meta. The companies are discussing a deal worth up to $10 billion over two years, with monthly payments and exit clauses for both sides. The proposal follows Anthropic's $45 billion compute agreement with SpaceX and comes as Meta plans to spend up to $145 billion on AI infrastructure in 2026. Leasing excess capacity could create a new revenue stream for Meta.

Kimi K3 tops coding benchmarks, outperforming Fable and GPT-5.6 Sol with frontier-level performance and a 1M-token context.

Alphabet shares fell 4% on July 16, 2026, after Bloomberg reported that Google's flagship Gemini 3.5 Pro AI model has been delayed by several months while engineers work to improve its performance. According to the report, the model's coding capabilities have not met internal expectations, placing Google under added pressure as rivals including OpenAI and Meta roll out stronger AI systems for software development.

Google said it is continuing to test Gemini 3.5 Pro with partners while emphasizing that it is shipping AI models quickly and cost-effectively. → Continue reading here.

Japan is partnering with NVIDIA and the newly formed Noetra consortium to build its first national AI factory, a 140-megawatt computing facility that will power the government's FRONTia program for robotics, industrial automation, and digital twins. The system will deploy 27,500 NVIDIA Rubin GPUs and 13,750 Vera CPUs across 382 Vera Rubin NVL72 racks, with funding beginning at ¥387.3 billion ($2.4 billion) and a long-term target of up to ¥1 trillion ($6.1 billion) through fiscal 2030.

The AI factory is designed to train large multimodal foundation models for robots and physical systems, with pretrained model weights made available to Japanese developers and businesses for industry-specific applications. The investment supports Japan's broader strategy to capture more than 30% of the global AI robotics market by 2040, positioning large-scale AI infrastructure as a national strategic asset. → Read the full article here.

China has introduced its first nationwide regulations targeting AI companion services, requiring chatbot providers to curb emotionally immersive interactions and banning virtual companion features for minors. Effective July 15, 2026, the rules also require companies to remind users they are interacting with AI, encourage breaks, and prevent emotional dependence that could harm real-world relationships.

Major firms including ByteDance and Alibaba have already begun removing personalized AI characters, even as AI companionship has become a lucrative business—accounting for 35% of MiniMax's revenue last year. The regulations underscore China's willingness to prioritize public welfare over rapid commercialization of AI, while still allowing domestic companies to market AI companion products internationally. → Read the full article here. (Paywall)

Netflix Expands Generative AI Use: Netflix says about 300 shows used generative AI this year, including for larger crowds and complex battle scenes.

AI Stock Sell-Off Spreads: Asian chip stocks slid after a U.S. tech sell-off, with SoftBank down 9% as investors questioned heavy AI spending.

Google Workers Seek Layoff Protections: More than 4,500 Google employees petitioned CEO Sundar Pichai for stronger layoff safeguards.

Apple Retakes Top Market Spot: Apple reclaimed the title of world's most valuable company, surpassing NVIDIA as investors shifted AI bets.

Databricks Targets $188B Valuation: Databricks is raising a funding round at a $188B valuation to expand its AI platform, research and acquisitions.

Adversarial clothing’: are garments designed to confuse facial recognition systems about to go mainstream? Designers say that as well as offering a degree of protection from surveillance, their clothes make a powerful fashion statement about the importance of privacy. Continue reading.

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team