Is 100 MW Enough for Data Center Demands?

One hundred megawatts of contracted power sounds substantial. It’s enough electricity to serve roughly 80,000 average American homes, committed through a long-term distributed power agreement to a single data center operator. When Kodiak's DPS subsidiary announced this arrangement as a centerpiece of its contract portfolio, it was framed as a signal of confidence in the distributed power model.

But here's the question worth asking: in an industry where hyperscalers are now designing individual campuses that consume 500 MW to 1 GW, is 100 MW actually a headline number — or a baseline?

The answer reveals something important about where data center energy strategy is headed and why the structure of power contracts matters almost as much as the megawatt total.

What Long-Term Power Contracts Actually Mean for Data Centers

A long-term power contract isn't just a procurement tool. It's a bet on the future — on where electricity prices are going, on grid reliability, on a facility's load trajectory, and increasingly, on how regulators and corporate sustainability commitments will shape energy sourcing decisions over the next decade.

For data centers specifically, these contracts typically take a few forms: traditional utility power purchase agreements (PPAs), direct contracts with generators (renewable or otherwise), or distributed power arrangements where power is generated and managed closer to the point of consumption. The DPS model that Kodiak has been building falls into that last category — and it's the most operationally interesting of the three.

Distributed power contracts offer something the standard utility relationship never could: contractual certainty combined with physical proximity to generation. When your power source is behind the meter or at the edge of the distribution network rather than dependent on 500 miles of transmission infrastructure, the risk profile changes entirely.

The trend toward these arrangements isn't accidental. Grid interconnection queues in the U.S. have swelled to over 2,600 GW of pending projects as of recent FERC data — meaning that getting new utility-scale power to a data center through traditional channels can take five to seven years. Long-term distributed power contracts are, in part, a workaround to a grid that simply wasn't designed for the scale of demand now being thrown at it.

The Real Scale of Data Center Energy Needs

To understand whether 100 MW is meaningful, you need a baseline for what modern data centers actually consume.

A small colocation facility might draw 5–20 MW. A mid-tier enterprise data center typically runs 20–100 MW. But the facilities being built by the hyperscalers — Amazon Web Services, Microsoft Azure, Google Cloud, and Meta — are operating in a different universe. Microsoft has announced data center campuses in Wisconsin and elsewhere targeting 150–300 MW per site. AWS is developing multi-site clusters in Northern Virginia where aggregate campus demand regularly exceeds 500 MW. The AI infrastructure wave has pushed this further: a single cluster of Nvidia H100 or H200 GPUs dense enough to train frontier models can consume 50–100 MW on its own.

The energy math for AI workloads is brutal — GPU clusters run at near-continuous high utilization, unlike traditional enterprise servers that might average 20–30% utilization rates. That changes the power density calculation completely.

So where does 100 MW land in this context? It's significant for a regional operator, a large enterprise data center, or a colocation provider serving multiple tenants. It's also meaningful for edge computing deployments or distributed AI inference facilities that don't require the concentrated density of a hyperscale training cluster. The DPS contract likely falls into one of these categories — and that's not a criticism. Not every data center needs to be a hyperscale behemoth, and the distributed power model may actually be better suited to mid-market operators who need reliability and cost predictability more than raw scale.

Why the Contract Structure Matters More Than the Megawatt Number

The operational benefits of long-term distributed power contracts are real, but they're often discussed in abstract terms. The specifics matter more.

Locking in power pricing over 10–20 years insulates a data center operator from the electricity price volatility that has blindsided facilities in deregulated markets. ERCOT in Texas, for example, saw spot prices spike to $9,000/MWh during Winter Storm Uri in 2021 — a risk that long-term contracted power largely sidesteps. For an operator running 100 MW continuously, the difference between $40/MWh contracted power and $80/MWh spot pricing over a year is roughly $35 million. That's not a rounding error. It's a business model.

Distributed power arrangements add a reliability dimension on top of the cost dimension. When generation assets are co-located or nearby, transmission risk drops. For facilities running latency-sensitive workloads or supporting critical infrastructure, that's worth paying a premium for — and increasingly, it's what enterprise customers are demanding in their SLAs.

The most sophisticated data center operators now view energy contracts not as a procurement function but as a core competitive differentiator. The ability to offer guaranteed uptime backed by contracted, proximate power is a sales argument that utility-dependent competitors can't easily replicate.

There are, however, legitimate risks to this structure. Long-term contracts are only as good as the counterparty backing them. If a distributed power provider faces financial distress, regulatory challenges, or operational failures, the data center operator is exposed — potentially more exposed than if they'd maintained a diversified utility relationship. Regulatory environments for distributed and behind-the-meter generation also vary significantly by state, and what's commercially viable in Texas or Arizona may face real headwinds in states with strong utility monopoly protections.

Challenges That Don't Show Up in the Press Release

The headline number of any power contract tends to obscure the operational complexity underneath it.

Capacity commitments are one thing; firm, dispatchable power is another. A 100 MW contract backed by intermittent renewable generation without adequate storage or backup creates a very different reliability profile than 100 MW of baseload gas or nuclear. Data centers — particularly those running AI workloads — need power that shows up on demand, not when the wind cooperates.

There's also the question of headroom. A data center operator signing a 100 MW contract today needs to model what their load looks like in years three, five, and ten. AI adoption is accelerating demand forecasts dramatically. Facilities that looked oversized two years ago are now running at capacity. Signing a contract that fits today's load without flexibility to expand could mean going back to an increasingly congested interconnection queue earlier than planned.

Regulatory considerations add another layer. Interconnection policy, net metering rules, and state-level renewable portfolio standards all affect the economics and feasibility of distributed power arrangements. The IRA's investment and production tax credits have improved the economics of renewable distributed power meaningfully, but the policy environment remains in flux — and a 15-year contract needs to survive multiple election cycles.

Where Data Center Energy Strategy Goes From Here

The Kodiak DPS deal is a useful data point precisely because it's not a hyperscale story. It's evidence that long-term distributed power contracts are becoming a standard tool across the market — not just for the companies with billion-dollar infrastructure budgets and armies of energy procurement specialists.

The next phase of this evolution will likely involve tighter integration between power contracts and storage assets. Battery storage paired with distributed generation fundamentally changes the reliability calculus — smoothing intermittency, providing frequency response, and allowing operators to optimize their draw from contracted versus spot market power in real time. Several data center developers are already structuring new facilities with co-located battery storage built into the project finance stack from day one.

Hydrogen and advanced nuclear are further out on the timeline, but both are attracting serious capital from data center operators who can see that the current grid simply cannot accommodate a future where every AI company, cloud provider, and enterprise IT department simultaneously tries to add hundreds of megawatts of load through traditional channels.

The operators who treat energy strategy as a core business function — rather than a facilities afterthought — will have a structural cost and reliability advantage that compounds over time.

One hundred megawatts, contracted through a distributed power arrangement with long-term pricing certainty, is a genuinely strong position for the right operator. The more important question isn't whether 100 MW is enough today. It's whether the contract structure gives that operator the flexibility, reliability, and optionality to still be competitively positioned in 2035. That's the due diligence question that belongs in every data center energy conversation — and it rarely gets asked loudly enough.

**Explore more about energy strategies for data centers at InfraSale Marketplace.**

[INTERNAL LINK: long-term power contracts]

[INTERNAL LINK: data center energy strategy]

[INTERNAL LINK: distributed power arrangements]