TL;DR

OpenAI revenue doubles; Microsoft, Meta raise AI CapEx; Nvidia faces China chip probe.

Highlights

• China’s cyber watchdog summoned Nvidia over alleged backdoors in H20 AI chips, demanding technical documentation and raising uncertainty for Nvidia’s China business.1
• OpenAI’s annualized revenue doubled to $12B, with ChatGPT exceeding 700M weekly users; the firm seeks an additional $30B in funding to support $8B in 2025 cash burn.2
• Microsoft reported strong Q4 earnings, with Azure sales up 39% and CapEx to exceed $30B; Meta raised 2025 AI CapEx guidance to $66–72B and outlined a new “personal superintelligence” initiative.3
• Meta’s Mark Zuckerberg signaled future frontier AI models may not be open-sourced, citing safety concerns; Meta Superintelligence Labs formed, led by former OpenAI scientist Shengjia Zhao.48
• References to GPT-5 surfaced in OpenAI and partner apps, with an early-August launch expected; OpenRouter’s “Horizon Alpha” model, possibly related to GPT-5, is now live for testing.56
• Google DeepMind launched AlphaEarth Foundations, an AI “virtual satellite” for 10-meter global mapping and environmental monitoring, integrated into Google Earth Engine.7
• Arm Holdings shares fell 8% after forecasting lower profits due to increased AI R&D and trade tensions; Arm confirmed plans to develop its own chips and expand in data centers.9
• Ambiq Micro surged 61% in its NYSE debut, highlighting investor demand for edge-AI hardware; its ultra-low-power chips are in over 270M devices.10
• UK’s CMA found Microsoft and AWS harming cloud competition, with potential antitrust action and stricter oversight under new digital market rules.11
• Google rolled out AI-powered “Age Assurance” in the US to restrict accounts for users under 18, including on YouTube.12
• Microsoft research identified 40 white-collar roles most exposed to AI automation, highlighting workforce reskilling needs.13
• Black Forest Labs and Krea released FLUX.1, an open-weight photorealistic text-to-image model, freely available for download.14
• WithNeo introduced “NEO,” a multi-agent AI system automating end-to-end machine-learning pipelines without human intervention.15

Commentary

AI infrastructure investment remains at the forefront, with Microsoft and Meta both committing to record CapEx for data center expansion as demand for AI workloads outpaces current capacity.34 Meta’s formation of Superintelligence Labs and its new focus on “personal superintelligence” signal a push for more advanced, user-facing AI systems, while also marking a strategic shift away from open-sourcing their most capable models.48 This move could tighten access to cutting-edge AI for the broader ecosystem, especially as the open/closed-source debate intensifies.

OpenAI’s rapid revenue growth and fundraising efforts underscore the commercial momentum behind generative AI, as ChatGPT’s user base continues to expand.2 The imminent release of GPT-5, with references surfacing in multiple partner apps and the appearance of “Horizon Alpha” on OpenRouter, suggests another step-change in model capabilities is near.56 Meanwhile, open-weight alternatives like FLUX.1 and automated ML platforms such as WithNeo’s NEO point to ongoing democratization of AI development, even as leading platforms consolidate market power.1415

On the hardware side, Nvidia faces renewed regulatory scrutiny in China over alleged chip backdoors, complicating its efforts to regain market share amid shifting US-China tech policy.1 Arm’s profit warning and strategic pivot toward in-house chip production reflect ongoing disruption in the semiconductor sector as AI workloads reshape demand from edge to data center.9 Ambiq Micro ’s strong IPO debut highlights continued investor interest in energy-efficient AI hardware for edge applications.10

Regulatory and compliance developments are accelerating. The UK’s CMA is preparing for possible antitrust action against Microsoft and AWS , and Google is deploying AI-powered age verification to address safety and regulatory requirements.1112 Microsoft ’s research on job exposure to AI automation further highlights the need for workforce adaptation as generative AI tools become more capable and widely adopted.13