Not All Load Is Equal
Why grid policy must distinguish between extractive and integrative compute
By Hailey Miller, Executive Director, Digital Power Network
The rapid growth of artificial intelligence, cloud computing, and digital infrastructure has forced an overdue conversation about the future of America’s electric grid. Policymakers are right to ask tough questions: Who pays for new generation? How do we protect households from rising electricity bills? How do we maintain reliability while decarbonizing?
But as this debate accelerates, a dangerous simplification is taking hold: the idea that all large electricity loads are the same.
They are not.
Treating every compute-intensive facility as interchangeable, whether hyperscale AI data centers or grid-interactive Bitcoin mining operations, will produce bad policy, higher costs, and worse climate outcomes. If we want a fair, reliable, and clean grid, we must learn to distinguish between extractive load and integrative load.
Not all load is equal.
The New Politics of Electricity Demand
For decades, electricity policy focused primarily on the supply side: generation, transmission, and fuel costs. Demand was treated as largely fixed. That era is over.
The explosion of compute-driven load has made electricity demand a central political issue. Regional grid operators are revising forecasts upward. Governors are demanding consumer protections. Federal agencies are exploring new rules to manage interconnection queues, pricing, and reliability.
These concerns are legitimate. Electricity is an essential service. Households should not be forced to subsidize unchecked industrial growth. Reliability must be protected. Climate goals must be met.
But none of these objectives can be achieved if policymakers fail to recognize that different types of load behave fundamentally differently on the grid.
The Category Error at the Heart of Today’s Debate
Much of today’s policy discussion treats “data centers,” “AI,” and “crypto mining” as a single category: large, energy-hungry, and inherently destabilizing.
This is a category error.
Traditional hyperscale data centers are designed to be always-on. They operate continuously, with minimal tolerance for interruption. Their value comes from uninterrupted service. Curtailment is expensive, disruptive, and often impossible.
Grid-interactive compute, such as many Bitcoin mining operations, works differently. These facilities are natively flexible. They can power down within seconds. They can absorb surplus generation. They can locate near stranded or underutilized energy resources. They can respond to price signals and grid stress in real time.
These distinctions are not academic. They are operational. And they matter for policy.
When the grid is stressed, some loads worsen the problem. Others help solve it.
Flexibility Is Not a Loophole, It Is Infrastructure
Flexibility is often discussed as a technical feature. In reality, it is a public good.
Flexible load can:
Reduce peak demand
Prevent blackouts
Stabilize prices
Enable higher renewable penetration
Make new generation financeable
This is why regional grid operators are now experimenting with policies that reward loads that bring or pair with generation. It is why regulators are exploring faster interconnection pathways for projects that reduce system risk. It is why long-term offtake contracts have become essential to building new clean energy.
These developments point to a crucial insight: the grid does not just need more supply. It needs smarter demand.
One-Size-Fits-All Rules Will Backfire
There is a growing impulse to impose uniform obligations on all large loads, special tariffs, long-term contracting requirements, capacity charges, or emissions standards.
The instinct is understandable. But the outcome would be counterproductive.
If flexible, grid-supportive loads are regulated as if they were inflexible baseload consumers, policymakers will discourage precisely the kinds of projects that can stabilize the system. They will slow clean energy development. They will increase, not reduce, system costs.
Worse, they will entrench the most rigid forms of consumption while punishing the most adaptive.
That is not fairness. That is misclassification.
The Climate Dimension
Climate policy cannot afford this mistake.
The fastest path to a low-carbon grid is not simply building more generation. It is aligning demand with clean supply. Flexible compute can absorb excess wind and solar. It can locate where renewables are abundant. It can help stabilize intermittent resources.
If policymakers fail to distinguish between grid-negative and grid-positive loads, they will inadvertently lock in more fossil generation, not less.
The irony is stark: rules meant to protect the climate could end up undermining it.
What a Smarter Framework Looks Like
The next generation of grid policy should be based on function, not form.
Instead of asking, “How big is this load?” we should ask:
Can it curtail?
Can it shift?
Can it co-locate with generation?
Can it provide grid services?
Does it reduce or increase system risk?
These questions lead to better outcomes for consumers, better outcomes for the grid, and better outcomes for the climate.
They also restore a core principle of public utility regulation: rates and rules should reflect how resources actually behave, not how they are labeled.
This Is a Moment of Choice
The politics of electricity demand are changing. For the first time in a generation, Americans are paying close attention to their power bills. Reliability is a top concern. The energy transition is accelerating.
In moments like this, bad categories produce bad policy.
We can choose a future where all large loads are treated as threats, regulated bluntly, and forced into rigid frameworks that raise costs and slow clean energy.
Or we can choose a future where the grid becomes more dynamic, more resilient, and more efficient, by recognizing that not all load is equal.
The grid does not just need more power. It needs better policy.