📨 AI’s Inflection Point: Power, Policy, Profit

This week's AI landscape reveals a critical inflection point: the race for technological dominance is colliding with mounting economic realities. The AI International Summit signaled Europe's strategic pivot from regulation to innovation while Trump's pushes to override regulation. Massive infrastructure investments by Microsoft and Middle Eastern sovereign funds underscore the global scramble for compute capacity. Yet Oracle's stock plunge and IBM's profitability warnings expose the financial strain beneath the AI buildout hype.

Today’s International AI Summit in Brussels, began by focusing on the strategic shift from regulation to innovation and industrial deployment. A standout theme: U.S.-China competition is forcing Europe to balance technological sovereignty with global collaboration. Panelists stressed that establishing interoperable standards is critical to prevent market fragmentation and preserve Europe's influence in AI governance. The discussion also surfaced practical adoption challenges: talent shortages, compliance complexity, and the risk of SMEs falling behind. Speakers emphasized that organizational readiness and cultural shifts matter as much as technological capability for successful AI integration.

President Trump announced plans for an executive order this week to preempt state AI regulation, arguing that companies need "one rulebook" rather than navigating 50 different state approval processes to maintain U.S. competitiveness against China. This follows a failed Senate attempt to insert AI preemption into the defense budget. The proposal faces bipartisan opposition, including from Republicans like Ron DeSantis and Marco Rubio, who argue it undermines federalism. Over 35 state attorneys general and 200 state lawmakers have warned against federal override, citing concerns about AI safety protections. [The New York Times]

Top Chinese firms (including Alibaba and ByteDance) are training AI models in Southeast Asian data centers to access Nvidia chips and circumvent U.S. export restrictions [Financial Times]. On Monday (Dec. 8), President Trump approved Nvidia's H200 exports to China with a 25% tariff for "approved customers" potentially restoring billions in revenue despite Democratic senators' national security warnings [Bloomberg]. Georgetown University professor Rush Doshi noted that "compute is our main advantage" over China, warning that "by giving this up we increase the odds the world runs on Chinese AI" [CNBC]. Meanwhile, Nvidia is developing location verification technology to combat chip smuggling, as the H200's processing performance is nearly 10 times previous export limits [Reuters].

Microsoft will invest $17.5B in India (2026-29) for AI and cloud infrastructure intensifying competition with Google and Amazon as India's massive digital user base becomes strategic for AI growth [TechCrunch]. Meanwhile, Brookfield and Qatar's sovereign wealth fund launched a $20B joint venture to build AI infrastructure in Qatar and international markets [Middle East AI News]. Both moves underscore the global race to secure compute capacity and regional AI dominance.

Oracle's stock plunged 11% after reporting a $15B CapEx surge for AI and cloud infrastructure, signaling squeezed near-term free cash flow and elevated leverage as hyperscalers race to meet demand amid doubts over ROI timelines [Forbes]. Meanwhile, IBM CEO Arvind Krishna warned that massive data center investments by Big Tech (potentially $1.5T for 20-30 GW capacity) defy profitability math due to soaring energy costs and infrastructure hurdles, with Goldman Sachs projecting AI power needs to hit 84 GW by 2027 [Fortune]. Both underscore the mounting financial strains and execution risks in the AI buildout, testing investor patience as capex balloons without immediate returns.

This week’s M&A highlights underscore the frantic push for AI infrastructure and regulated compliance. Major deals show Big Tech securing real-time data flows, while strategic acquisitions target the specialized tools needed for model development and managing regulatory risk.

This week's selection highlights ventures building foundational AI capabilities necessary for global competition, focusing on solving the economic risks of the AI buildout, autonomous execution for efficiency, and the critical technological frontier of embodied intelligence at the frontier of robotics.

This week's selection for growth investment opportunities focuses on companies that are scaling proven solutions to address the most urgent operational risks cited earlier: AI-driven compliance and regulatory complexity, autonomous enterprise efficiency (AI Agents), and the paramount need for next-generation cybersecurity in an interconnected, AI-accelerated world.

The European Commission opened an antitrust investigation into Google over concerns the company is using publishers' online content and YouTube videos for AI training and AI Overviews without sufficient consent or adequate compensation. Regulators worry Google is abusing its search dominance to impose unfair conditions on publishers. Google faces potential fines up to 10% of global annual revenue if found guilty of violating EU antitrust rules. [Reuters]

AI agent startups attracted over $2.8B in H1 2025 as 93% of IT leaders plan implementation by year-end 2026, driving a 275% surge in Agentic AI earnings call mentions. However, Gartner (global research and advisory firm) forecasts 40% of projects will fail by 2027 due to escalating costs, unclear ROI, and inadequate risk controls. Early enterprise deployments show promise in constrained applications (coding assistants and specialized research) with some achieving 30% efficiency gains. Critical barriers remain: compounding language model errors undermine reliability in multi-step tasks, while security vulnerabilities expose systems to data theft and manipulation. 🔗 Read the report by Thoughtworks and WIRED

Researchers demonstrated parallel optical matrix-matrix multiplication (POMMM) using coherent light propagation, achieving performance consistency with GPUs while offering advantages in bandwidth, parallelism, and energy efficiency. The method performs complex tensor operations (critical for neural networks) in a single light pass, addressing scalability limitations of existing optical computing approaches. The team demonstrates compatibility with convolutional and vision transformer networks, with potential for multi-wavelength expansion and integration with existing photonic platforms, positioning optical computing as a viable path for next-generation AI infrastructure. [Read the article published in Nature Photonics]

📅 16-20 Feb 2026 | 🌐 New Delhi

Flagship summit bringing together policymakers, industry leaders, researchers, startups, civil society, and international organizations to advance responsible AI, governance, and innovation for inclusive growth & sustainability — featuring 300+ exhibitors, keynotes from Nvidia's Jensen Huang, Anthropic's Dario Amodei, Reliance's Mukesh Ambani, and more, plus research symposia and global dialogues on People, Planet, and Progress. 🔗 Register

That’s it for this week.

Until next time,

The CLNM Capital

[Written and edited with the assistance of AI]

Want to share this newsletter with a friend? Use this link: https://clnm.beehiiv.com/