I’ve been watching the energy side of the hyperscaler story for a while now, and every so often a development lands that forces you to sit up a little straighter. This past week delivered one of those moments. A company that has spent months trying to lock down its first major customer finally put a name to the mystery tenant everyone had been whispering about. The reaction in the market was immediate and loud. Shares jumped hard on the news, then settled into a still-impressive gain. Yet underneath the celebration sits a quieter, more complicated reality: the nuclear half of the plan keeps slipping further into the future.
Amazon Steps Back In With Serious Commitment
On Thursday the developer behind the project announced that an unnamed investment-grade hyperscaler had committed to an initial 600 megawatts starting in 2027. The deal came with a $200 million prepayment and a smaller early construction agreement worth $15 million. By Friday the identity was no longer a secret. The customer is Amazon. The stock responded the way you would expect when a big, creditworthy name finally signs on the dotted line. It climbed as much as 42 percent at one point before closing the day still up more than a quarter.
This feels less like a brand-new relationship and more like a return engagement. Earlier reporting had linked Amazon to the project, only for those stories to be denied. A previous mystery tenant had walked away from a substantial construction advance, and the fallout was painful. Now the same name is back, this time with real money and a clearer timeline. In my view that prepayment matters almost as much as the megawatts. It gives the company breathing room and a concrete signal that the customer is serious.
What the New Lease Actually Replaces
Interesting detail that got less attention than the Amazon headline: the new agreement replaces rather than stacks on top of an earlier deal. A previous lease covering 222 megawatts with a different customer, valued in the billions, had already missed its closing deadline. That arrangement has now been terminated, and the credit support that came with it is gone. So the company is not adding Amazon on top of a solid book of business. It is swapping one commitment for another. The net effect is still positive because of the quality of the new counterparty, but the path has not been linear.
Hardware progress offers a more tangible bright spot. The latest update shows all six of one turbine model already on site, five of six units from another major supplier also present, and additional machines in transit. Roughly two gigawatts of generation capacity has been secured in one form or another. A service agreement with a regional utility adds the possibility of up to 200 megawatts of grid power as a bridge. Mobilization for the new customer is targeted for the beginning of next month, with first power expected sometime in the second half of 2027. Up to 925 megawatts could be available that year, 600 of it reserved for Amazon.
The gas-fired head start is real and useful. It does not, however, restore the years that have already slipped away on the nuclear side of the ledger.
Boardroom Drama Continues in the Background
While the commercial team was busy closing the Amazon deal, the boardroom has remained lively. The former chief executive, who was dismissed earlier this year, has continued to push for a sale of the company and has pursued legal action against the firm and certain directors. He managed to score a symbolic win at the most recent annual meeting when holders aligned with him voted heavily against the re-election of a high-profile director. That director and another fell short of the majority required under the bylaws, yet both remain in place as holdovers. The company maintains that the vote violated an earlier nomination agreement. These internal tensions have been public for months and show little sign of quieting down completely.
From an outsider’s perspective the ongoing friction is distracting, though not necessarily fatal. Investors tend to care most about signed contracts and visible construction progress. Still, repeated public fights rarely help when the company is trying to attract additional large customers or keep regulators comfortable.
Nuclear Timelines Keep Moving to the Right
Here is where the story gets less cheerful. The long-term vision has always included a substantial nuclear component built around four large reactors of a proven design. The schedule for those units has now slipped again. The company currently expects to bring one reactor online each year beginning in 2034 and running through 2037. Earlier guidance had pointed to a start in 2033. The original ambition was even more aggressive, targeting the first unit closer to the beginning of the next decade.
Most of the delay appears tied to the licensing process. The final package of required information that must be submitted for full regulatory review has been pushed out by roughly ten months. An earlier target of the end of 2026 has become late 2027. The remainder of the updated application, which incorporates the latest design revisions from the reactor vendor, is now expected in the middle of 2027. These are not minor administrative adjustments. They push the entire nuclear program further into the future and increase the period during which the company will rely primarily on gas-fired generation and purchased power.
Leadership on the nuclear side has also changed. The executive who had been overseeing the nuclear construction effort left the company in early October. Correspondence with regulators that once went to him now goes to the chief operating officer. Continuity matters in these long, complex licensing processes, and any change at the top introduces at least some risk of further delay or lost institutional knowledge.
Not the Only Project Facing Slips
It is worth noting that Fermi is hardly alone in watching nuclear timelines stretch. Other advanced reactor projects, including one of the more closely watched small modular reactor efforts in North America, have also acknowledged delays and cost pressures. First-of-a-kind engineering challenges, supply chain realities, and regulatory thoroughness all play a role. In this case the technology is not first-of-a-kind; versions of the same reactor design are already operating elsewhere. That fact makes the slip more notable, not less.
The practical consequence is straightforward. Amazon’s commitment and the associated prepayment give the company a solid gas-fired foundation and a path to meaningful revenue in 2027. They do not accelerate the nuclear portion of the plan. The two-year loss relative to earlier expectations remains on the books.
Hardware Arriving While Paperwork Lags
One of the more encouraging pieces of recent disclosure concerns physical progress on the gas side. Turbines are not just ordered; many of them are already sitting on the site or en route. That is the kind of tangible evidence investors like to see after years of announcements and shifting dates. The company has also lined up a utility partnership that can supply interim power while its own generation ramps up. These are practical steps that reduce execution risk in the near term.
Contrast that with the nuclear paperwork. The gap between physical readiness on the gas units and regulatory readiness on the reactors is wide and growing. I have found that markets often reward visible steel in the ground more quickly than they punish distant licensing delays, at least until those delays start affecting the overall equity story. For now the Amazon news has dominated the narrative.
- All six units of one major turbine model are already on site
- Five of six units from a second supplier have arrived
- Additional machines are in transit
- Approximately two gigawatts of generation capacity has been secured
- A utility agreement provides a bridge of up to 200 megawatts
Market Reaction and What It Signals
The sharp move in the shares tells you that the market had been waiting for exactly this kind of customer validation. After previous disappointments, including a tenant walk-away and management upheaval, confidence had eroded. A recognized hyperscaler willing to prepay a substantial sum restores some of that confidence overnight. Whether the enthusiasm proves durable will depend on execution over the next eighteen to twenty-four months. First power in 2027 is no longer a distant concept; it is close enough that missed milestones will be harder to excuse.
Perhaps the most interesting aspect is how cleanly the gas strategy now stands on its own. The nuclear ambition remains part of the long-term story, yet the company can generate meaningful revenue and demonstrate operational capability well before the first reactor comes online. That sequencing may prove more realistic than earlier plans that tried to keep both tracks moving in tighter parallel.
Looking Ahead to the Next Milestones
Several near-term markers will matter. Mobilization for the Amazon load is scheduled to begin very soon. Progress reports on turbine installation and interconnection work will be watched closely. On the nuclear side, the quality and completeness of the materials submitted to regulators in 2027 will set the tone for the remainder of the decade. Any further slippage there will be measured against the original promise of earlier commercial operation.
Internal governance issues are unlikely to vanish overnight. The former chief executive continues to hold a significant stake and has shown willingness to use every available lever. How the board and remaining management navigate that relationship while still delivering for customers will be another subplot worth following.
In the broader context of data-center power demand, this episode is a microcosm of a larger tension. Hyperscalers need enormous amounts of reliable electricity on relatively short timelines. Gas can be delivered faster. Nuclear offers longer-term advantages on emissions and fuel costs but carries heavier regulatory and construction timelines. Companies that can bridge the two, as this project is attempting to do, will attract attention. Execution, not ambition, will decide which ones succeed.
Balancing Near-Term Revenue Against Long-Term Vision
The $200 million prepayment is more than symbolic. It provides working capital at a moment when the company needs to move from planning into heavier construction spending. It also locks in a high-quality customer for a meaningful block of capacity. Those are real positives. At the same time, the repeated movement of nuclear commercial operation dates to the right means the carbon-free portion of the portfolio arrives later than once hoped. Investors who bought the original story for its nuclear content now have a longer wait.
I tend to view the current setup as pragmatic. Secure the near-term load with technology that can be deployed on a commercial timeline. Keep the nuclear program alive and progressing through the licensing maze. Accept that the two tracks will not stay perfectly synchronized. That approach may lack the elegance of earlier presentations, yet it looks more achievable given the evidence so far.
| Item | Earlier Expectation | Current Outlook |
| First nuclear unit | Around 2032-2033 | 2034 |
| Subsequent units | Annual thereafter | One per year through 2037 |
| License package completion | End of 2026 | Late 2027 |
| Gas-fired first power | Earlier targets missed | Second half of 2027 |
Why the Prepayment Changes the Risk Profile
Cash in the door from a creditworthy customer reduces financing risk and gives management more options. It also signals to other potential tenants that the project is moving from concept toward reality. In a sector where many announcements remain paper-thin for years, tangible prepayments stand out. The earlier departure of a different tenant had raised questions about the company’s ability to close. This deal answers some of those questions, at least for the gas phase of the project.
Still, one large customer does not make a fully contracted campus. The company will need to continue filling capacity beyond the initial 600 megawatts if it wants to support the full build-out. Success with Amazon could make those conversations easier. Failure to deliver on time would make them harder. The next two years will be revealing.
The Human Element Behind the Delays
Projects of this scale are never just about equipment and permits. People leave. Priorities shift. Regulatory staff ask for more information. Vendors revise designs. Each of those events is understandable in isolation. Cumulatively they move schedules. The departure of the nuclear construction lead is one more data point in that pattern. Whether the transition proves seamless will become clearer as the revised application materials take shape next year.
On the commercial side the company appears more stable. Securing Amazon after earlier setbacks suggests the remaining team can close deals. That capability may ultimately matter more in the medium term than perfect continuity on the nuclear org chart.
Broader Implications for Hyperscaler Power Strategies
Data-center operators face a simple arithmetic problem. Demand for computing capacity is rising faster than new generation can be built in many regions. Waiting exclusively for new nuclear or large-scale renewables plus storage leaves them short. Contracting for gas-fired power that can be online in a few years, while keeping longer-term nuclear or other firm low-carbon options in the pipeline, has become a common hybrid approach. This project is one example of that pattern playing out in real time.
The market’s enthusiastic response to the Amazon news suggests investors understand the logic. They are willing to underwrite the gas bridge as long as the customer quality is high and the near-term milestones look credible. The nuclear delays are noted, priced in to some degree, and set aside for later evaluation.
What Success Looks Like from Here
Success in the next phase is relatively easy to define. Mobilize on schedule. Install and commission the turbines without major surprises. Deliver first power to Amazon in the targeted window. Continue making steady progress on the nuclear license application so that 2027 submissions are complete and high quality. Manage internal governance issues so they do not distract from operations. If those boxes get checked, the stock’s recent strength may prove justified. If any of them slip significantly, the celebration will look premature.
I have watched enough energy development stories to know that the gap between announcement and electrons is where most of the risk lives. This company has now narrowed that gap on the gas side and widened it on the nuclear side. The Amazon commitment buys time and credibility. It does not eliminate the harder work still ahead.
The coming quarters will show whether the organization can convert a strong customer win into consistent operational delivery. That, more than any single headline, will determine how the longer story is written.
In the end the week delivered both validation and a reminder. Validation that a major hyperscaler sees enough substance to commit serious capital. A reminder that nuclear schedules remain stubbornly difficult to compress. For anyone following the intersection of digital infrastructure and large-scale power, both pieces of information are useful. The gas turbines are arriving. The reactors will take longer. The customer is real. The next test is execution.
