This week’s AI developments highlighted a shift from model hype to the realities of infrastructure, data, and regulation. Apple’s decision to use Google’s Gemini for future AI features, Meta’s push to lock down massive compute and energy capacity and Microsoft’s effort to ease community resistance to data centers all pointed to AI scaling becoming as much a political and logistical challenge as a technical one.
At the same time, scrutiny intensified around how AI is built and deployed. OpenAI’s reported effort to source real-world work samples raised new IP questions, regulators moved against xAI over misuse concerns and fresh research suggested humanoid robotics remains limited by physical constraints, not intelligence. Meanwhile, big capital continued to flow into AI infrastructure, chips and robotics.
Industry & Enterprise
Apple Selects Google Gemini to Power Next Phase of AI Features Across Its Platforms
Apple confirmed a multi-year, non-exclusive partnership with Google to use Google’s Gemini models and cloud infrastructure to support future AI features, including a long-anticipated upgrade to Siri, following internal evaluations of multiple providers. Apple said core privacy safeguards will remain in place with significant processing kept on-device, while the deal—reportedly valued around $1 billion—unfolds amid ongoing U.S. antitrust scrutiny of Google’s default and exclusivity arrangements. (AI Insider)
OpenAI Reportedly Expands Training Data Efforts by Requesting Real-World Work Samples From Contractors
OpenAI is asking third-party contractors to upload examples of real work from past and current jobs to generate higher-quality training data for its models, according to a report by Wired, as part of an effort carried out with data provider Handshake AI. Legal experts cited by Wired warned the approach could raise intellectual property risks, noting that the process relies heavily on contractors’ judgment to determine what material is confidential or proprietary, potentially increasing exposure as AI training operations scale. (AI Insider)
Meta Launches Meta Compute to Scale AI Infrastructure and Expand Global Energy Capacity
Meta launched Meta Compute to accelerate expansion of its AI infrastructure, signaling a push to build tens of gigawatts of capacity this decade—and potentially hundreds over time—as large-scale AI workloads drive energy and compute demand. The initiative, overseen by infrastructure chief Santosh Janardhan, Daniel Gross, and Dina Powell McCormick, positions Meta alongside Microsoft and Alphabet in a tightening race to secure AI-ready data centers, networks, and energy supply. (AI Insider)
Anthropic Appoints Irina Ghose to Lead India Expansion Ahead of Bengaluru Office Launch
Anthropic has appointed Irina Ghose, former Managing Director of Microsoft India, to lead its India business as it prepares to open an office in Bengaluru, signaling a push to make India a core growth market. India is already Anthropic’s second-largest market by users for its Claude models, and the company plans to expand enterprise, developer, and startup adoption by building local teams and partnerships. (AI Insider)
Google Defends AI Shopping Protocol After Consumer Group Raises Pricing Concerns
Google has pushed back against criticism of its newly announced Universal Commerce Protocol after consumer advocates, including Groundwork Collaborative, warned the AI-powered shopping framework could enable personalized price manipulation. Google said merchants cannot display prices higher than their own websites and that its AI shopping agents cannot alter prices based on individual user profiles, as scrutiny grows over how AI-driven commerce uses consumer data and incentives. (AI Insider)
Microsoft Unveils 5-Point Plan for ‘Community-First AI Infrastructure’
Microsoft launched a “Community-First AI Infrastructure” initiative to address rising local opposition to data center expansion, committing to measures spanning electricity pricing, water use, job training, property taxes, and community investment as AI infrastructure spending accelerates. In a blog post, Vice Chair and President Brad Smith said large-scale AI buildouts will stall unless communities see clear, measurable benefits that outweigh pressures on power, water, and local services. (AI Insider)
Research & Innovation
What Are The Remaining Bottlenecks For Humanoid Robotics? A Physical AI Study Finds The Limits Are Physical, Not Cognitive
A new Physical AI study finds that progress in humanoid robotics is constrained less by model capability than by data scarcity, sim-to-real gaps, energy limits, and the difficulty of safely coordinating whole-body physical interaction. The research concludes that advances in simulation fidelity, energy-efficient hardware, real-time control, and safety verification—not larger AI models—will be critical to achieving reliable performance outside controlled environments. (AI Insider)
Policy & Governance
California Attorney General Opens Investigation Into xAI’s Grok Over Sexualized AI Imagery
xAI is facing escalating regulatory scrutiny after Rob Bonta opened an investigation into its chatbot Grok over the generation of nonconsensual sexually explicit images, following reports that users on X prompted the system to create sexualized images of real individuals, including minors. The probe comes amid parallel actions by regulators in the European Union, United Kingdom, India, Indonesia, and Malaysia, as xAI begins implementing limited safeguards while authorities assess potential violations of laws governing nonconsensual sexual imagery and child sexual abuse material. (AI Insider)
Startups & Capital
European Commission Commits €307M to Advance Trustworthy AI and Strategic Digital Technologies
European Commission has opened two new funding calls under its Horizon Europe Digital, Industry and Space programme, committing €307.3 million to advance artificial intelligence and emerging digital technologies across the bloc. The package includes €221.8 million for trustworthy AI, data services, and Apply AI-aligned areas such as robotics and quantum technologies, plus €85.5 million focused on open strategic autonomy, next-generation AI agents, industrial robotics, and advanced sensing materials, as part of the EU’s broader competitiveness agenda. (AI Insider)
United States and Taiwan Announce $250B Investment Pact to Expand AI and Semiconductor Manufacturing
U.S. Department of Commerce announced a trade and investment agreement with Taiwan committing Taiwanese semiconductor and technology firms to invest $250 billion in U.S. chip manufacturing and AI-related production, with an additional $250 billion in credit guarantees to support expansion. The deal, finalized under the Trump administration, aims to reshore advanced semiconductor capacity critical to AI and national security as the U.S. moves to reduce reliance on foreign supply chains following new tariffs on advanced AI chips. (AI Insider)
OpenAI and SoftBank Commit $1B to SB Energy to Expand AI Data Center Infrastructure
OpenAI and SoftBank plan to invest $1 billion in SB Energy as part of a strategic partnership tied to the $500 billion Stargate initiative unveiled with Oracle last January, with SB Energy set to build and operate OpenAI’s 1.2-gigawatt data center in Milam County, Texas. Backed by SoftBank and Ares Management, SB Energy will also enter a non-exclusive preferred partnership to develop a repeatable model for large-scale AI data center construction as OpenAI scales compute capacity to meet rising demand. (AI Insider)
Andreessen Horowitz Raises Over $15B Across New Funds to Back AI and Strategic Technologies
Andreessen Horowitz has raised more than $15 billion across five new funds, including $6.75 billion for growth investments, $1.7 billion for AI infrastructure, and $1.12 billion focused on national interest areas such as defense and supply chains, underscoring continued investor appetite for AI-driven strategies. The raise, completed less than two years after its prior fund cycle, brings the firm’s assets under management to more than $90 billion and highlights how large, established venture firms continue to draw capital despite a broader slowdown in venture fundraising. (AI Insider)
Skild AI Raises $1.4B, Now Valued Over $14B
Skild AI has raised nearly $1.4 billion in funding led by SoftBank Group, valuing the company at more than $14 billion and ranking among the largest raises to date for a robotics-focused AI company. The round included NVentures, Bezos Expeditions, Macquarie Capital, and new strategic backers such as Samsung, LG, Schneider Electric, CommonSpirit Health, and Salesforce Ventures, as Skild AI scales its omni-bodied robotics foundation model across enterprise environments without hardware-specific retraining. (AI Insider)
Dassault Aviation Leads $200M Series B in Harmattan AI to Advance Autonomous Combat Systems
Dassault Aviation has formed a strategic partnership with Harmattan AI and is leading the company’s $200 million Series B to accelerate controlled autonomy and AI integration across next-generation combat aviation systems. The funding will support Harmattan AI’s global scale-up following Programs of Record with the French and UK defense ministries, expanding AI-enabled missions and industrial production across ISR, drone interception, and electronic warfare in alignment with Dassault’s Rafale F5 and UCAS roadmaps. (AI Insider)
Torq Secures $140M Series D at $1.2B Valuation to Lead the AI SOC and Agentic AI Era
Torq raised $140 million in a Series D led by Merlin Ventures, valuing the company at $1.2 billion and bringing total funding to $332 million as adoption of its AI-driven security operations platform expands. The company said the new capital will support go-to-market growth, including U.S. federal and public sector expansion, and scale autonomous SOC operations globally as enterprises automate alert triage, investigations, and response. (AI Insider)
Check out AI Insider every day for the latest developments in AI and robotics.
