TL;DR

OpenAI plans trillion-dollar data center spend; Nvidia wins partial China chip nod; Meta faces Senate AI probe.

Highlights

• OpenAI CEO Sam Altman plans to raise and spend “trillions” on global data center expansion, warning of an AI investment bubble but projecting positive long-term impact 1.
• U.S. government is in talks to take an equity stake in Intel to boost domestic chip manufacturing, aiming to strengthen U.S. semiconductor supply chains 3.
• Nvidia receives U.S. approval to export downgraded AI chips to China with a new 15% sales tax, while major funds increase Nvidia holdings; China has not yet accepted the terms 2.
• AI data centers now account for 4% of U.S. power use (25% in Northern Virginia), with projections of 12% by 2028 and similar surges in Taiwan, driving up consumer electricity bills and prompting regulatory action 814.
• Galaxy Digital raises $1.4B to convert its Texas Helios site into a large-scale AI datacenter, supplying up to 800 MW to CoreWeave 7.
• Google DeepMind launches Gemma 3 270M, a 270M-parameter language model optimized for smartphones and edge devices, available for commercial use with safety restrictions 4.
• Meta faces bipartisan Senate scrutiny after internal documents reveal AI chatbots could engage in inappropriate conversations with children and provide unsafe advice; company is reviewing standards 5.
• OpenAI considers introducing advertising to ChatGPT’s 700 million weekly users as it seeks new revenue streams and responds to user backlash over model changes 9.
• Lambda (Nvidia-backed) and Rivos seek major funding rounds ($4–5B and $2B+ valuations, respectively) as AI chip and infrastructure competition intensifies; Nvidia affirms next-gen GPU timeline 6.
• MIT and University of Pennsylvania researchers use AI to design novel antibiotics, with MIT’s model generating two candidates effective against drug-resistant gonorrhoea and MRSA 11.
• Eli Lilly signs a $1.3B AI drug discovery deal with Superluminal Medicines for obesity and cardiometabolic therapies 12.
• Grab invests in WeRide to deploy Level-4 robotaxis across Southeast Asia, integrating autonomous tech with ride-hailing and targeting phased commercial rollout 13.
• China opens the first World Humanoid Robot Games, featuring 500 robots from 16 countries, signaling ambitions in embodied AI 15.
• Stanford researchers demonstrate a brain-computer interface decoding inner speech with 74% accuracy, advancing neurotech for paralyzed patients and raising privacy questions 10.

Commentary

The AI sector’s capital requirements and operational footprint continue to expand rapidly, with OpenAI’s multi-trillion-dollar data center ambitions 1 and Galaxy Digital ’s $1.4B Texas buildout 7 highlighting the scale of infrastructure investment underway. This growth is driving significant increases in electricity demand—already 4% of U.S. consumption and up to 25% in key regions 814—resulting in higher residential bills and new regulatory measures 8. Similar trends in Taiwan and federal actions in the U.S. underscore energy supply as a strategic constraint for AI scaling 14.

On the hardware front, U.S.-China technology tensions remain central. Nvidia ’s partial approval to export downgraded chips to China, subject to a 15% tax, keeps its China business in flux, even as institutional investors increase exposure 2. The U.S. government’s potential equity stake in Intel signals a more direct approach to securing domestic semiconductor capacity 3. Meanwhile, Lambda and Rivos are aggressively raising capital, reflecting sustained investor appetite for AI chip and cloud infrastructure startups 6.

Product and research announcements remain strong. Google DeepMind ’s Gemma 3 270M targets edge AI with efficient on-device performance and broad commercial licensing 4. In healthcare, MIT’s AI-designed antibiotics 11 and Eli Lilly ’s $1.3B partnership with Superluminal Medicines for AI-driven drug discovery 12 show continued momentum in AI-enabled R&D. Stanford’s inner speech BCI breakthrough points to practical advances in neurotechnology, though privacy remains a concern 10.

Regulatory and ethical scrutiny is intensifying. Meta ’s internal policy lapses around AI chatbot safety have triggered bipartisan calls for federal regulation 5, while OpenAI’s consideration of advertising for ChatGPT reflects ongoing pressure to balance monetization with user trust 9. Autonomous mobility is also advancing, with Grab and WeRide moving toward large-scale robotaxi deployments in Southeast Asia 13 and China spotlighting robotics ambitions with its World Humanoid Robot Games 15.