TL;DR

U.S. takes 10% Intel stake; Apple weighs Gemini for Siri; Google lands $10B Meta AI cloud deal.

Highlights

• The U.S. government has taken a 10% non-voting equity stake in Intel via CHIPS Act grants; similar stakes in Nvidia and Apple are proposed under a $700B Trump administration plan to tie subsidies to equity13.
• Apple is in talks with Google to license Gemini for a revamped Siri, evaluating Gemini, in-house, and other third-party LLMs2.
• Google has started training a Gemini variant for Apple’s Private Cloud Compute and secured a $10B, six-year Google Cloud deal with Meta for AI infrastructure5.
• Meta licensed Midjourney’s image/video generation tech to enhance its generative AI models and consumer products4.
• Google claims a 33-fold drop in Gemini’s AI query energy use and published new methodology for industry-wide sustainability metrics6.
• OpenAI released Codex-CLI 0.23, increasing usage limits by 50%, adding GPT-5 integration, and supporting image input7.
• Google opened Veo 3 AI video generation to all Gemini users for a free weekend trial, previously limited to paid subscribers8.
• Google Cloud launched a sub-$1/year “Gemini for Government” plan for U.S. federal agencies, undercutting OpenAI and Anthropic10.
• Waymo received New York City’s first permit to test robotaxis on public roads, expanding its autonomous vehicle operations9.
• Lightning AI launched a Multi-Cloud GPU Marketplace, enabling up to 70% cost savings by routing workloads across providers11.
• Higgsfield released Kling 2.1 for advanced video editing and WAN 2.2 with one-click deepfake presets, raising content safety concerns1314.
• xAI announced “Macrohard,” aiming to automate software development entirely with AI agents, challenging traditional developer roles12.
• AI-powered voice restoration enabled a woman with motor neurone disease to speak again after 25 years15.

Commentary

Government involvement in the U.S. semiconductor sector is intensifying, with the administration converting CHIPS Act subsidies into equity stakes in Intel and proposing similar moves for Nvidia and Apple 13. This approach directly links public funding to financial returns and aims to incentivize domestic investment, but it also raises questions about government influence in corporate strategy and market dynamics for AI hardware supply chains.

On the platform side, Google is deepening its AI partnerships and infrastructure reach. The company is training a custom Gemini model for Apple’s Private Cloud Compute, potentially powering a major Siri overhaul, and has secured a $10B multi-year cloud and AI infrastructure contract with Meta 5. These moves, combined with Google ’s sub-$1/year Gemini for Government plan, signal a push to expand market share in both enterprise and public sector AI workloads by leveraging scale, custom silicon, and aggressive pricing10.

The competitive landscape in generative AI and developer tools continues to evolve rapidly. Meta ’s licensing of Midjourney’s imaging tech4, OpenAI’s expanded Codex-CLI with GPT-57, and Google ’s Veo 3 trial8 reflect ongoing efforts to differentiate on model quality, multimodal capabilities, and accessibility. Meanwhile, Lightning AI’s GPU marketplace addresses ongoing compute constraints11, and Higgsfield’s new video features highlight both creative opportunities and emerging content safety risks1314.

Autonomous systems and AI’s role in healthcare also made notable advances. Waymo’s approval to operate robotaxis in New York City marks a significant regulatory milestone for urban AV deployment9. Separately, AI-powered voice restoration for neurodegenerative disease patients underscores the technology’s growing impact in assistive applications15.