Anissa Gardizy

Anthropic is pursuing its first data center leases to control its own server infrastructure for AI development, aiming to reduce computing costs long-term. The company has signed more than a dozen letters of intent with U.S. data center developers. Anthropic is also in discussions with Google about a financial guarantee arrangement for lease payments, with Google potentially co-designing server chips for the facilities.

Broadcom announced a new financing fund backed by Apollo and Blackstone to fund more than 20 gigawatts of AI data centers through 2028 using Broadcom-designed chips. The fund will support infrastructure projects tied to Anthropic and OpenAI. Apollo is leading an initial $35 billion commitment to the effort.

OpenAI is in advanced negotiations to lease a 10 gigawatt data center campus on federal land in Ohio, with potential financial backing from Nvidia. The facility would be among the largest of its kind, with a total build-out cost of at least $500 billion at current prices for chips, labor, power, and materials. OpenAI would control equipment under a long-term lease and assume payment obligations once operations begin, with the first phase expected in 2028.

Data center developer Switch is in advanced fundraising talks to raise billions at a valuation exceeding $50 billion, according to reporting from The Information. The round involves investors including Brookfield Asset Management and KKR. A deal at this valuation would position Switch among the most valuable privately held data center operators.

Barkr, a Miami fintech startup founded in 2023, is building financial infrastructure to value GPUs as loan collateral by providing insurance-backed valuations that lenders can rely on. The company addresses a critical gap in AI infrastructure financing: while GPUs have become a major asset class, lenders and borrowers often disagree on residual values, making GPU-backed loans less favorable than traditional asset-backed deals. In six months, Barkr has facilitated roughly $200 million in GPU transactions and expects to handle $300 million more by year-end, helping standardize what has been a fragmented and uncertain market.

Google and Blackstone are launching a joint cloud computing venture to rent Google's tensor processing units (TPUs) to AI developers. The partnership represents Google's effort to expand TPU availability beyond its own cloud platform and compete more directly with other AI infrastructure providers. The new entity will operate as a dedicated TPU cloud provider, targeting developers and companies that need specialized hardware for training and running AI models.

Modal, a startup providing GPU rental and software infrastructure for running and training AI models and agents, is in talks to raise funding at a $4.5 billion valuation, representing an 80% jump from its previous round just months ago. The valuation surge follows a five-fold increase in annualized revenue to approximately $300 million since last fall, driven primarily by demand for Modal's sandbox environments that allow developers to run AI agents and code in isolated software spaces without affecting their broader systems.

The Huang Foundation, the charitable organization of Nvidia CEO Jensen Huang and his wife Lori, has signed a deal to rent Nvidia GPUs from CoreWeave with the intention of donating them to AI developers. The arrangement, disclosed in Nvidia's annual report, represents a structured approach to philanthropic GPU distribution in the AI ecosystem. The foundation has already committed $108 million toward this initiative, signaling a significant capital allocation toward supporting AI research and development outside Nvidia's direct commercial channels.

Anthropic has signed a seven-year cloud infrastructure deal worth $1.8 billion with Akamai, the content delivery network provider. The contract was announced by Akamai on Thursday, and the revelation that Anthropic was the unnamed customer behind the deal drove Akamai's stock up 27% on Friday. The agreement represents a significant commitment to cloud capacity and signals Anthropic's infrastructure scaling needs as it expands its AI services.

OpenAI and Broadcom announced a custom AI chip partnership last fall positioned as finalized, with plans to deploy enough capacity to consume 10 gigawatts of power by 2030 and reduce reliance on Nvidia. However, the companies have not resolved how OpenAI will finance the estimated $18 billion project, creating a significant execution risk for a deal that was publicly presented as already locked in.

Google announced plans to sell TPUs directly to customers for deployment in their own data centers, marking a shift away from Google Cloud exclusivity. However, executives from three major neocloud providers, Nebius, Lambda, and CoreWeave, indicated they have no near-term plans to adopt TPUs, citing overwhelming market demand for Nvidia GPUs and the concentrated nature of TPU adoption. Google has responded by pivoting toward a narrower strategy, striking a deal with Fluidstack to deliver TPUs to Anthropic rather than attempting broad ecosystem adoption across multiple neocloud platforms.

Anthropic is in talks with London-based startup Fractile to purchase inference chips designed to run AI models efficiently, with availability expected in 2026. The move would diversify Anthropic's chip supply beyond current partners Google, Amazon, and Nvidia as the company's server demands surge alongside explosive sales growth. Securing additional suppliers could give Anthropic negotiating leverage as its annual spending on servers and chips approaches tens of billions of dollars.

Kos.AI, a San Francisco startup, raised $12 million to build software that automates the back-office accounting work for AI data center developers. The company addresses a critical pain point: as companies prepare to spend trillions on data center infrastructure, finance teams are overwhelmed reviewing massive invoices (often 800+ pages) and contracts (thousands of pages) from general contractors billing $500 million monthly. Kos functions as a virtual employee, connecting to email, Slack, and enterprise systems to review invoices, verify details, and track purchase orders, with human oversight built in.

Phononic, a 17-year-old Durham, North Carolina semiconductor company that makes cooling components for AI data center servers, is in talks with potential buyers at a valuation of at least $1.5 billion, with some buyers expressing interest above $2 billion. The company has engaged investment bank Lazard to evaluate its options since early 2026. This valuation would more than double its last private funding round, reflecting broader investor appetite for industrial suppliers tied to AI infrastructure demand. Phononic may also choose to raise additional capital instead of pursuing a sale.

OpenAI has committed to spending more than $20 billion over three years on servers powered by Cerebras chips, a significant move to reduce reliance on Nvidia hardware and potentially lower computing costs. As part of the deal, OpenAI will receive warrants for a minority stake in Cerebras with the option to increase ownership as spending grows. OpenAI is also providing approximately $1 billion to fund data center development for Cerebras infrastructure that will support its AI products.