Every market cycle produces a listing that forces investors to redraw their categories. For the London Stock Exchange’s real estate segment, the current candidate is Fermi Inc (LSE:FRMI) — a Texas-based company that calls itself a REIT, builds power plants alongside data centres, and arrived on the London market in late 2025 with gigawatt-scale ambitions and barely any revenue.

Fermi is unlike anything else carrying a London real estate quote. Co-founded by former US Energy Secretary and Texas governor Rick Perry, the company is developing Project Matador, a vast campus near Amarillo in the Texas Panhandle designed to combine private electricity generation with hyperscale data centres serving artificial-intelligence workloads.

The scale of both ambition and valuation is striking. The company has set out plans for a multi-gigawatt private grid — with reports citing targets of 11 gigawatts and beyond over the coming decade — and has said it is on track to deliver about 1.1 gigawatts of new power by the end of 2026. Recent market data put the value of its London line at roughly £3.85 billion, with analyst consensus reported as Buy.

For UK investors accustomed to REITs that collect rent from warehouses or offices, Fermi poses a genuinely novel question: how do you price a development-stage landlord whose product is powered land for the AI age?

Project Matador: Power First, Data Centres Second

Fermi Inc is a US real estate investment trust, dual-listed on Nasdaq and the London Stock Exchange under the ticker FRMI, that is developing what it describes as a hyperscale energy and AI data centre campus in Texas.

The company’s founding insight is that electricity, not land or fibre, has become the binding constraint on AI infrastructure. Waiting times for grid connections in the United States stretch for years, while AI developers need power in gigawatt quantities on short timescales. Fermi’s answer — a “power-first” model — is to build its own private, off-grid generation and wrap data centre capacity around it.

Project Matador sits on a site in Carson County, adjacent to the US Department of Energy’s Pantex facility, that the company has called one of the most strategically valuable pieces of undeveloped land in America for large-scale energy and computing infrastructure. Plans span natural-gas turbines initially, with solar, batteries and, over the longer term, nuclear ambitions layered on top, alongside as much as 18 million square feet of data centre space.

Progress markers have come quickly for a young company. According to company statements and filings, Fermi expects to deliver roughly 1.1 gigawatts of power at the campus by the end of 2026, with a stated path towards double-digit gigawatts by the late 2030s. In February 2026 it secured a US$500 million loan from Mitsubishi UFJ Financial Group, and press reports in 2026 suggested that OpenAI had been evaluating capacity at Project Matador — sending the shares sharply higher, though no definitive agreement had been confirmed at the time of writing.

The REIT structure is deliberate: Fermi intends to earn long-term, lease-like revenues from tenants occupying its powered shells, translating an energy-infrastructure project into a real estate income model that institutional investors recognise.

A London Quote on the AI Power Race

Fermi’s presence on the LSE says as much about the exchange as the company. London’s real estate segment has spent years dominated by domestic landlords trading at discounts to asset value; a US-based, development-stage AI infrastructure REIT choosing a London line in late 2025 gave the market rare exposure to the theme that has dominated global equities.

It is important to be clear about what Fermi is not. It is not a UK property company: it owns no British assets, earns nothing in sterling and has no connection to the UK commercial property cycle beyond its listing venue. Investors screening it alongside conventional LSE property shares are comparing fundamentally different propositions.

The demand backdrop, however, explains the enthusiasm. Forecasts across the industry point to extraordinary growth in electricity consumption from AI data centres, and hyperscalers have been signing multi-gigawatt agreements with power developers on terms unimaginable five years ago. Data centre capacity with secured energy has become one of the scarcest commodities in global real estate.

Fermi competes in that race with a differentiated claim: speed. By building behind-the-meter generation rather than queueing for grid connections, it argues it can deliver powered capacity years faster than rivals. Its proximity to natural-gas supplies in the Texas Panhandle and its stated efforts on water efficiency — the company has told Texas lawmakers Project Matador would cut water use substantially versus conventional cooling — form part of that pitch.

The competitive set is formidable, though: utilities, hyperscalers building their own campuses, and established data centre REITs are all chasing the same tenants, and the sector’s economics will ultimately be set by supply as well as demand.

Why Investors May Be Watching Fermi Inc

The bull case rests on scarcity and scale. If AI compute demand continues on anything like its current trajectory, entities that control large blocks of near-term power capacity could command premium economics, and Fermi’s 2026 delivery target of about 1.1 gigawatts would place it among the earlier movers at genuine scale.

Credibility signals have accumulated. A US$500 million facility from one of the world’s largest banks, a dual Nasdaq-LSE listing, high-profile leadership and reported interest from frontier AI developers together suggest the project is being taken seriously by counterparties whose diligence matters.

The REIT wrapper is part of the attraction for some investors: if Fermi succeeds in signing hyperscale tenants to long leases, today’s speculative development story could evolve into a long-duration income stream — the same journey, in compressed time, that logistics landlords made over the previous decade.

Analyst sentiment, with consensus recently reported as Buy, reflects that optionality. So does the share register’s appetite for news: the stock has moved violently on reports such as the possible OpenAI evaluation, behaviour more typical of growth technology names than of real estate investment trusts.

For UK-based followers of real estate income investing, Fermi is effectively a call option on the AI build-out, wearing a REIT’s clothing — a characterisation that captures both the appeal and the danger.

Key Factors to Monitor

• Tenant announcements: signed leases or capacity agreements with hyperscalers or AI developers would be the single most important validation of the model; reports of interest are not contracts.

• Power delivery milestones: whether the company hits its stated target of roughly 1.1 gigawatts online by the end of 2026, and the pace of subsequent phases.

• Funding events: further debt or equity raises to finance a multi-billion-dollar build-out, and the dilution or leverage they imply.

• Construction execution: turbine deliveries, supply-chain bottlenecks and build costs in an overheated market for generation equipment.

• Regulatory and permitting matters: air, water and land permissions in Texas, plus any federal developments affecting off-grid generation or nuclear ambitions.

• Competitive dynamics: the volume of rival powered-capacity announcements, which will shape pricing when leases are eventually struck.

• Path to revenue: the transition from development-stage spending to recurring, REIT-style rental income and, eventually, any dividend policy.

Risks to Consider

Fermi is a speculative, development-stage company, and the risks deserve unusual emphasis for a stock classified under real estate. It is largely pre-revenue: today’s multi-billion-pound valuation rests almost entirely on execution that has yet to happen, and any slippage in power delivery, tenant signings or funding could trigger sharp derating.

Concentration risk is extreme. Essentially all value sits in one campus in one Texas county, exposing shareholders to local permitting, water and community issues, weather events and the fortunes of a single construction programme.

The demand thesis itself could disappoint. If AI infrastructure spending slows — because model efficiency improves, capital tightens or hyperscalers over-build — the projected scarcity of powered capacity could ease, undermining the economics of speculative campuses everywhere.

Financing is a standing question: gigawatt-scale energy and data centre construction consumes billions, and reliance on continued access to debt and equity markets leaves the company vulnerable to funding-condition shifts. Equipment lead times for turbines and electrical gear add further execution hazard.

UK investors face additional layers: dollar-sterling currency exposure, a secondary London line whose liquidity may lag Nasdaq, no dividend, and volatility more akin to a technology growth stock than a property trust. Reports of prospective tenants should be treated as unconfirmed until the company announces binding agreements.

Outlook: The Milestones That Could Define FRMI’s Year

The next twelve months are unusually legible for a company this young: the market may judge Fermi almost entirely on whether first power arrives on schedule and whether reported tenant interest converts into signed capacity agreements.

A confirmed hyperscale lease could transform the story from concept to contracted cash flows; continued silence, or slipped timelines, might test the patience embedded in the valuation. Funding announcements will be read closely for terms as much as size.

Investors may also watch how the broader AI power race evolves — rival campuses, utility responses and any regulatory shifts in Texas — since Fermi’s premium rests on staying ahead of a fast-moving field. Nothing about this trajectory is assured, in either direction.

Conclusion

Fermi Inc has imported Silicon Valley-scale ambition into the London real estate list: a power-first, off-grid AI campus REIT targeting gigawatts where conventional landlords count square feet. The prize, if Project Matador delivers and tenants sign, is a claim on the defining infrastructure build-out of the decade. The risks — pre-revenue status, single-site concentration, funding hunger and an unproven model — are equally outsized. Among new names on the LSE, LSE:FRMI may well be the sector’s most watched, and its most polarising.