On May 1, 2026, the Wall Street Journal reported that Coatue — the firm straddling venture capital and hedge funds — has launched a venture called Next Frontier. The strategy compresses to one sentence: buy parcels of land near large power sources and convert them into data centers. This is not Coatue’s first AI bet — it already holds sizable stakes in Anthropic, OpenAI, xAI, CoreWeave, and Singapore’s DayOne — but this time it is not buying equity. It is buying the genuinely scarce inputs of the AI build-out: land and power.
The context explains the timing. According to Pew Research, the United States has roughly 3,000 data centers today with more than 1,500 additional ones at some stage of construction, and most new sites are landing in rural areas. Compute demand is growing considerably faster than the grid and qualified land can be unlocked.
What Next Frontier Actually Does
The business model the Journal describes is not complicated: find land next to power infrastructure, buy it, and wait for it to become a data center. The value comes from scarcity — parcels with electricity, water, fiber, and a viable grid interconnection are a limited set, and whoever holds them sits upstream in every later negotiation. A hedge fund can also move on land faster than a hyperscaler’s procurement process, then monetize that optionality through leases, joint ventures, or outright sales once the site is qualified. For Coatue, it is a way to upgrade its AI exposure from passive equity stakes to active control of a critical input, with a higher ceiling on returns. Coatue declined to comment when TechCrunch reached out.
The Fluidstack JV and the Anthropic Thread
The most concrete development in the report is that Next Frontier has already signed a joint venture with Fluidstack, a cloud infrastructure startup. Fluidstack is not an obscure name: it previously struck a $50 billion deal to build data centers for Anthropic, and in mid-April it was reported to be in talks for a $1 billion round at an $18 billion valuation — only months after its previous round valued it at $7.5 billion.
That chain gives the “possibly for Anthropic” framing its substance. The joint venture partner’s anchor customer is Anthropic, and Anthropic itself is reported to be raising at roughly a $900 billion valuation that could close within two weeks (we examined those talks earlier). If Next Frontier land ends up serving Fluidstack’s Anthropic projects, Coatue would hold exposure at three layers of the stack simultaneously — the model company, the operator, and the ground underneath.
Land Near Power: The New Speculative Asset
Coatue is hardly alone in seeing the trade. Blackstone is involved in financing Related Digital’s $16 billion Oracle data center project in Saline Township, Michigan, and Shark Tank investor Kevin O’Leary is pitching his own data center land financing play. The wave of capital started earlier, too: in January, Nvidia invested $2 billion in debt-laden CoreWeave to help it add 5 gigawatts of AI compute.
When chipmakers, private equity, venture funds, and celebrity capital all converge on land-plus-power, the signal is clear: the bottleneck of AI infrastructure has moved upstream to siting and interconnection. Whoever controls qualified land controls the pace at which the next tranche of compute comes online. For data center developers, the implication is stark: the cheapest compute in three years may belong to whoever locked qualified sites in 2026, not to whoever orders the most GPUs.
The Risk Side of the Land Trade
Land speculation is not risk-free arbitrage. First, interconnection and permitting move on year-long timelines, and carrying costs eat into returns the whole time. Second, if AI demand growth slows, or model efficiency improves faster than expected, land acquired in a frenzy becomes a stranded asset. Third, pricing in this market is still supported by a thin set of comparable transactions, so valuations rest on a shaky foundation. There is also a political dimension: rural communities are already weighing the jobs and tax revenue data centers bring against grid, water, and land-use impacts, and local opposition can stall even a qualified site indefinitely. Coatue’s thesis requires the compute shortage to persist until these parcels are developed — which is exactly where market consensus currently sits, and precisely why it deserves scrutiny.
Sources
- Coatue has a plan to buy up land for data centers, possibly for Anthropic — TechCrunch
- AI Investor Coatue Joins Data-Center Frenzy With New Venture to Buy Land — The Wall Street Journal
- Most new data centers in the U.S. are coming to rural areas — Pew Research Center
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.