Hut 8’s (HUT) transaction with Anthropic (ANTHRO) and FluidStack to provide artificial intelligence data center capacity, which features a financial backstop with Google (GOOG)(GOOGL), is a “new high water mark for deal economics,” according to Morgan Stanley.
As part of the deal, Hut 8 will develop and deliver at least 245 megawatts and up to 2.3 gigawatts of AI data center infrastructure for Anthropic, leveraging high-performance clusters operated by Fluidstack. The partnership is structured across three tranches.
“We view this transaction as representing a new high water mark with respect to economics provided to Bitcoin players who are building a powered shell for a data center customer,” said Morgan Stanley analysts, led by Stephen Byrd, in a Thursday investor note.
Morgan Stanley said the deal is indicative of value creation of $17 to $19 per watt.
“We’re able to retain the relationship, the upside of demand from Anthropic, Fluidstack and Google, but be able to have an investment-grade counterparty be able to provide a guarantee and financial backstop on all lease payments, which is about $7 billion for the term of the first 15 years, minimum power bills, which is about another $1 billion, property tax, insurance and other pass-through expenses as well,” said Hut 8 CEO and co-founder Asher Genoot during a call with investors on Wednesday.
In the first phase, the partnership will begin at Hut 8’s River Bend campus in Louisiana, where Hut 8 and Fluidstack plan to build 245 MW of IT capacity, supported by 330 MW of utility power. Under tranche 2, Hut 8 plans to grant the tenant (Fluidstack) a right of first offer for up to 1,000 MW of additional IT capacity at River Bend, subject to the expansion of power at the site. Under the final tranche, Hut 8 and Anthropic may jointly develop up to 1,050 MW of additional optional capacity to scale across Hut 8’s development pipeline beyond River Bend.
“We are seeing improving terms for power and data center development transactions,” Byrd noted. “Power and data center developers may be able to achieve improved terms, such as the allocation of project risks, given the significant excess of demand for these products relative to the supply (and this mismatch is growing, in our view). This is a helpful development, because we see significant project execution risk, ranging from supply chain to labor challenges.”
“We are fundamentally bullish on the magnitude of benefits from AI adoption, and we expect this magnitude to increase rapidly as LLMs continue to improve in 2026 at a non-linear rate,” Morgan Stanley added.