Learning From the Past
Christine Primmer, a Senior Advisor with Clarum Advisors, is an investor/operator with diverse expertise helping venture-backed technology organizations build scale and resilience. She’s closely advised, operated, and partnered with twenty-plus venture-backed tech companies in energy, transportation, smart buildings, and IOT, ranging from Series A to E, helping them grow revenue and scale operations using data driven strategies, partnerships/M&A, and capital.
Everyone is talking about AI and robotics. Much speculation exists about the implications — from deepfakes to the future of humanity itself. For the power sector and its stakeholders, the issue boils down to two truths:
AI leadership will be a defining geopolitical race this century.
AI-driven data centers are pushing load growth at a pace and concentration the grid hasn’t seen in decades.
Why This Matters
The utility model was constructed to deliver one promise: safe, reliable, affordable power for all — from your grandmother in Iowa to the power-hungry factory in Georgia. This stability has powered economic mobility and national competitiveness for the last century.
The AI iron rush presents new challenges for this paradigm. Powering it represents the same existential importance as electrifying homes in the early 1900s. But it needs to be done responsibly, with consideration for:
Need for Speed: balancing hyperscaler timelines with regulatory oversight.
Customer Affordability: preventing outsized bill impacts to average consumers.
Climate Risk — or Opportunity: meeting demand fast with gas could raise emissions; meeting it with clean capacity could accelerate decarbonization.
Physical and Regulatory Bottlenecks: data center developers face constraints not just in power access, but also zoning, permitting, and water availability — each adding friction to timelines and siting.
Utilities are all-hands-on-deck — rate cases, incremental upgrades, tech pilots. But revolutionary and structural innovation is needed in addition to effort. The current model will be challenged to scale at AI speed.
Learning from Historical Precedents
This isn’t the first time incumbents have faced demand spikes they couldn’t finance or deliver alone. Two models stand out:
TowerCos in telecom: In the 2000s, independent firms built and owned cell towers, leasing them back to carriers. Beyond the financial incentive of moving assets off carrier balance sheets, TowerCos unlocked shared asset economics — multiple carriers could co-locate on common infrastructure, driving massive scale and efficiency.
Independent Power Producers (IPPs) and merchant solar developers: Over the past two decades, developers carried permitting, financing, and construction risk, then sold site-ready projects via PPAs and tax equity. Utilities gained clean power without the upfront burden, while markets saw a rapid expansion of privately financed generation capacity.
Key Lessons
Specialization equals speed. Creating roles for players who focus solely on building and financing assets can unlock velocity.
Risk and capital allocation matter. When entities with different risk appetites share the load, projects move faster without jeopardizing balance sheet stability.
Partnership, not replacement. New models complemented incumbents — they didn’t erase them.
A Working Theory
We can’t copy-paste TowerCos or IPPs into today’s challenge. Telecom infrastructure cycles faster, and renewables benefited from strong tax policy tailwinds. Meanwhile, AI-driven load growth is geographically concentrated and physically resource-intensive.
But the principle holds: structural innovation — alongside technology — may be the unlock.
Part 2 will map the emerging archetypes — from hyperscalers to infra funds — and ask: will they complement utilities, or compete with them?
