The AI race has found its bankers—and they may end up owning the part that matters. Here’s how we got here: Anthropic had already locked in a major European compute deal—a $10 billion, six-year commitment to Volta Infra for 121 IT megawatts of Nvidia Vera Rubin capacity at Bitdeer’s Tydal campus in Norway. Backed by a credit backstop, that deal had already pushed the Claude maker toward a wider, multi-provider supply strategy before this latest infrastructure platform. This is AI Daily Briefing. Today, we’re asking: who pays for the concrete, who signs away the capacity, and who gets the upside if inference gets cheap fast? Here's Georgie Preston at Investor Daily:
On 11 August, the asset manager said it had partnered with GIC and Anthropic to launch Theseus Infrastructure, a new platform that will develop, operate and lease data centre infrastructure to Anthropic under long-term agreements. Funds managed by Macquarie Asset Management, along with GIC, will own the platform and fund the majority of each project.
Macquarie and GIC own Theseus Infrastructure; Anthropic is the anchor tenant on long leases. It’s a REIT-shaped way to handle the AI compute bill: institutions own the concrete, Claude rents it. And this is a real platform, launched August 11, with dedicated U.S. sites to develop and operate—not another glossy capacity announcement. Anthropic’s compute supply strategy keeps widening: after the Norway capacity deal, Macquarie and GIC are building dedicated sites. I like the contracted-demand piece more than the usual speculative build. But if Anthropic locks into leases at today’s compute economics and inference gets cheaper fast— somebody has to absorb that mismatch. Tenant, landlord, or both. Macquarie and Singapore’s GIC are funding most of these projects. That tells you how far this market has moved: sovereign and infrastructure capital are now directly financing model labs’ infrastructure. Anthropic also says it’ll cover consumer electricity-price increases from these sites. Put that commitment in the lease documents—and keep it there. A $122 billion fundraise sounds less like venture capital and more like a national infrastructure project. So what is OpenAI actually buying with that money—and how do we tell if this is a business investment rather than an arms race? Computing capacity: chips, data centers, and the electricity needed to run increasingly costly AI systems. Reuters reports that OpenAI raised $122 billion at an $852 billion valuation. CNBC’s Ashley Capoot and Kate Rooney reported that the company told investors it was targeting roughly $600 billion in total compute spending through 2030. CNBC also reported that OpenAI was aiming for about $280 billion in 2030 revenue after bringing in $13.1 billion the prior year. Investors are effectively underwriting a huge gap between today’s sales and future infrastructure needs. And the wider buildout is enormous: Reuters columnist Karen Kwok wrote that Alphabet, Amazon, Meta, and Microsoft planned about $650 billion in spending this year, largely on chip-filled data centers that also serve cash-burning labs. Much of this funding is meant to secure access to AI’s physical machinery before rivals do. Revenue targets are easy to put in a pitch deck. What would actually show that the economics are improving? Watch whether recurring revenue grows fast enough to support the compute commitments, and whether OpenAI can keep lowering the cost of delivering its products as usage expands. CNBC reported that CFO Sarah Friar told employees annualized recurring revenue in July exceeded all of the second quarter. That’s a useful operating signal, but eventually that growth has to narrow the gap with infrastructure spending. For builders and investors, the test is whether demand can pay for the power-hungry capacity being built to serve it. If you’re enjoying AI Daily Briefing, please subscribe or leave us a review wherever you’re listening. Reviews help other listeners find the show, and your support helps us keep bringing you the day’s essential AI news.
Links to every story we covered today are in the show notes if you’d like to follow up on the ones that caught your attention. Thanks for listening, and we’ll be back tomorrow. That’s AI Daily Briefing for today. This is a Lantern Podcast.