Capital flowing to physical and operational layers necessary if AI systems are to scale.
The latest artificial intelligence-related transactions show that the center of gravity in AI deal‑making has continued its decisive shift toward infrastructure — power, data centers, optics, and security — rather than consumer applications or standalone models. Capital is flowing to the physical and operational layers that determine whether AI systems can scale at all.
For our running list of hundreds of deals check the Mogin Law A.I. Deal Table. Also, read our previous update: AI “Chokepoint” Deals Are Reshaping the Competitive Map of the Tech Stack
BlackRock / Nvidia / Microsoft consortium acquires Aligned Data Centers
A BlackRock‑led group, backed by Nvidia and Microsoft through the Artificial Intelligence Infrastructure Partnership, agreed to acquire Aligned Data Centers from Macquarie in a deal valued at roughly $40 billion. Aligned operates more than 50 data center campuses with over five gigawatts of operational and planned capacity across the Americas. The transaction is expected to close in the first half of 2026. Source: Yahoo! Finance.
Why it matters: This is one of the largest pure‑play AI infrastructure transactions ever announced and reflects the growing view that AI‑ready data centers are strategic assets akin to energy or transportation infrastructure.
BlackRock / EQT‑led consortium to acquire AES Corp
A consortium led by BlackRock’s Global Infrastructure Partners and Swedish private‑equity firm EQT agreed to acquire U.S. power company AES Corp in a transaction valued at approximately $33.4 billion, including debt. Reuters and analysts explicitly linked the deal to surging electricity demand from AI data centers, which is straining U.S. power grids and accelerating investment in reliable generation and transmission assets. The transaction is expected to close in late 2026 or early 2027. Sources: Reuters, New Straits Times, U.S. News & World Report.
Why it matters: This is one of the clearest examples yet of AI demand reshaping “old‑economy” sectors. Power generation is emerging as a strategic choke point for AI growth, drawing infrastructure‑scale capital into utilities.
AMD disclosed a long‑term strategic partnership with Meta covering roughly six gigawatts of AI infrastructure capacity. The agreement includes semi‑custom GPUs and multi‑generation collaboration, signaling Meta’s effort to diversify its AI compute stack beyond Nvidia while securing long‑duration capacity for training and inference workloads at hyperscale. Source: Yahoo! Finance
Why it matters: The deal underscores how hyperscalers are locking in compute years in advance and reducing single‑vendor risk as AI infrastructure becomes both scarcer and more capital‑intensive.
Nvidia–Lumentum strategic optics partnership
Nvidia announced a multiyear strategic partnership with Lumentum to develop advanced optical interconnect technologies for next‑generation AI data centers. The nonexclusive agreement includes multibillion‑dollar purchase commitments and a $2 billion Nvidia investment to expand Lumentum’s U.S. manufacturing capacity. The focus is on solving data‑movement bottlenecks inside massive AI clusters rather than on models or software. Sources: Business Insider, Business Times.
Why it matters: As GPU counts scale into the hundreds of thousands, optical interconnects are becoming as critical as compute. This deal highlights the shift from “AI models” to “AI factories” as the dominant investment frame.
Nvidia–Coherent optics partnership
Alongside Lumentum, Nvidia also announced a parallel multiyear optics partnership with Coherent, including a $2 billion equity investment and multibillion‑dollar purchase commitments. The agreement focuses on silicon photonics, laser components, and high‑bandwidth optical networking for large‑scale AI deployments. Source: Business Insider.
Why it matters: Nvidia is effectively vertically integrating the AI supply chain beyond chips, locking up future capacity in critical photonics components that could otherwise become bottlenecks.
OpenAI acquires Promptfoo
OpenAI agreed to acquire Promptfoo, a startup specializing in AI model testing, red‑teaming, and security evaluation, with plans to integrate Promptfoo’s tools directly into OpenAI Frontier, its enterprise platform for building and operating AI agents. OpenAI described the deal as a move to make security, compliance, and behavioral testing “native” to enterprise AI deployments. Financial terms were not disclosed. Sources: CRN.com and Bloomberg.
Why it matters: As AI agents move into regulated, high‑risk workflows, security tooling is shifting from a bolt‑on feature to core infrastructure. This acquisition positions OpenAI to compete not just on model quality, but on enterprise‑grade governance and risk controls.
AI Infrastructure Capital Flood Continues
February closed with a record $189 billion in global AI venture funding, driven overwhelmingly by infrastructure and foundation‑model companies. The largest rounds included OpenAI’s $110 billion raise and Anthropic’s $30 billion financing, underscoring the accelerating concentration of capital at the top of the AI stack and intensifying questions around compute access, scale advantages, and vertical integration. Source: Crunchbase News (Mar. 3, 2026).
Why it matters:
These financings confirm AI’s shift from speculative technology to core economic infrastructure, likely increasing antitrust scrutiny of market power, cloud exclusivity arrangements, and the risk of downstream foreclosure as dominant players control both models and compute.
