Data Centers over People
How we can learn from the current Tahoe situation to be prepared in Southern Arizona
This is an opinion piece on recent developments in the US Data Center landscape and potential impacts on the Southern Arizona Region.
This is something I was worried we would see, but I was optimistic that the public fallout from a decision like this would have been enough to dissuade anyone from taking the action. For those who haven’t seen the news, on Saturday (March 20, 2026), CalMatters (a non-partisan non-profit newsroom covering California) posted the following story: Nevada utility to Lake Tahoe: Find electricity elsewhere
The story is worth the read, and while much still needs to be sorted out, it appears that the primary electricity supplier for Liberty Utilities, which serves Tahoe, NV Energy, has opted to focus its current electrical generation capacity on data center customers rather than continuing to power the town, leaving the utility until May of 2027 to find a new supplier. While NV Energy’s comment in the story is that they are aware of the situation and are reviewing it, at least as presented, an energy provider has opted to prioritize data center power consumption over actual human beings.
I will say this is something that has been in the back of my mind regarding the Marana Data Center projects, given the power generation gaps here in Southern Arizona relative to these projects. TRICO is primarily a power purchaser, which potentially puts them in a similar position as Liberty Utilities. If their suppliers can do better with their production capacity, and they sell directly to data centers, could we end up with a similar tangled web, leaving residents on the outside?
I developed a thought experiment around the decision, if it were mine to make, and I boiled it down to five areas: economics, risk, operational fit, strategic optionality, and regulatory/political exposure.
Economics
Hyperscale data centers offer structurally superior unit economics. They sign long-term PPAs (often 10–20 years) at negotiated rates with near-100% capacity factors. A 750 MW facility, like one of the proposed here, running at 90%+ load factor is extraordinarily valuable to a generator. Municipalities, by contrast, are served under regulated rate structures with set margins and regulatory gates. Revenue per MWh is lower, and cost recovery timelines are longer. The rules in Arizona allow infrastructure cost recovery in the rate formula, which can be a hidden financing advantage.
Risk
The risks point in multiple directions. Hyperscaler counterparty credit is excellent when it is the direct end users, such as Microsoft, Google, and Amazon. They are among the highest-rated corporate obligors in the world. It gets murkier when private equity comes into play, such as Blue Owl, which has been the subject of significant reporting about its potential liquidity strain (CNBC’s most recent article). Customer concentration risk is severe: one contract, one customer, one termination clause. If the hyperscaler renegotiates, exits, or the facility doesn’t get built, the generator is left with stranded capacity. Municipal contracts carry political and regulatory risks (rate cases, franchise disputes, city council changes) but also exhibit systemic stickiness; the timeframe to lose a municipal customer is slow and drawn out, and the franchise relationship creates durable revenue.
Operational Fit
This is where a hyperscale data center is genuinely advantageous from a generator standpoint. Data center load is flat, constant, and predictable. Municipal load is volatile: morning ramps, summer AC peaks, winter troughs, and in Southern Arizona, these can be severe. Serving municipal load requires investment in peaking capacity, demand response infrastructure, and ancillary services. Serving a hyperscaler requires sustained baseload reliability and near-zero downtime tolerance, an expensive, but plannable proposition.
Strategic Optionality
Municipal franchise relationships often include exclusivity within a service territory, creating a long-term moat. Hyperscale relationships, while lucrative, are more transactional and at risk of hyperscalers building their own generation assets.
Regulatory and Political Exposure
Data centers in the Southwest face significant and growing regulatory scrutiny, air quality permitting (RICE NESHAP, Title V), water use, grid interconnection queues, and local land use approval. There is significant exposure to the permitting risk for customer projects, especially as communities become more active in resisting these projects and moratoriums become more commonplace. Municipal service is a regulated construct, slower to change, politically exposed, but backed by the state utility regulatory framework.
The most critical variable is whether the power generators are regulated utilities or merchant generators. Regulated utilities have a fiduciary obligation to serve municipal load, and they can't simply redirect capacity. Merchant generators have full optionality, and the hyperscale case is a substantially cleaner financial story.
TRICO is a partial-requirements member of Arizona Electric Power Cooperative (AEPCO) and receives most of its energy and capacity from AEPCO. AEPCO is a not-for-profit generation and transmission (G&T) cooperative providing wholesale power to six member distribution cooperatives in southern Arizona. AEPCO is not a merchant generator. It operates at cost-of-service for the benefit of its member co-ops (including TRICO), not to maximize margin in a competitive wholesale market. Now, whether it has the will/desire to expand its capacity for these consumers if its members request is a separate question, but at least the existing supply, in its current form, appears protected.
TRICO would need to procure new supply to support the Marana Data Center project, likely through new AEPCO contracts, PPAs with third-party generators, or its own new generation assets. That procurement would flow through the cooperative governance and potentially ACC review, rather than be a unilateral commercial decision, which is crucial. It provides members with visibility and a say in the process, both of which are essential. It becomes essential that we monitor the process from TRICO, identify the third-party generators, and categorize them, because that is the essential exposure here.
TEP is a separate issue at the moment, and with the new Franchise Agreement in flux, understanding it in detail is essential; the outcome of the vote (November 2026 I believe) is critical.
While none of this is a guarantee one way or the other, the thought experiment is at least valuable for exploring where potential holes might exist, enabling a decision like the one that appears to be happening in Tahoe to happen here in Southern Arizona. As Seneca said, "luck is what happens when preparation meets opportunity," and in this case, if we understand how these decisions are made and the processes involved, we know what we can do to influence or counter them.