Inside Digital Infrastructure: Luminary Strategies’ Arushi Sharma Frank on Rethinking Risk, Power, and AI Infrastructure

Step into the full digital infrastructure ecosystem with our Q&A series featuring innovators who are redefining what’s possible across data centers, fiber networks, broadband, and next‑generation connectivity. Hear from investors, lenders, developers, and operators on what’s changing across power, capital, and risk — and what it takes to scale at speed.

Arushi Sharma Frank, Founder and Principal of Luminary Strategies, joins Steven Shparber, Co-chair of Mintz’s Digital Infrastructure Practice, to discuss one of the defining challenges facing the AI economy: how to connect rapidly growing demand for power with a grid that was not built to accommodate it at today’s scale. Drawing on her work shaping market reforms in ERCOT, Arushi explains why Speed to Power is fundamentally a risk-allocation challenge, how flexible interconnection frameworks can accelerate infrastructure deployment without compromising reliability, and why the most successful developers and investors will be those that learn to underwrite and manage power-delivery risk themselves.

Why Large AI Loads Require a New Approach to Risk, Reliability, and Grid Participation

Steven: How should the market think about large AI loads: primarily as demand growth to be managed, or as flexible infrastructure that can help support the grid?

Arushi: Large AI loads are unquestionably demand growth. But the more important question is who underwrites the risk of serving that demand.

The grid is shared infrastructure. It cannot simply become the underwriter of every private return associated with an enormous new load. Large power users should come to the grid with tools that make the shared system stronger: dispatchability, batteries, generation, computational response, controls, and contracts that allow them to manage more of their own power-delivery risk.

So I would move beyond the word “flexibility.” What matters is credible private underwriting. A large load becomes valuable infrastructure when the grid can see it, dispatch it, rely on its response, and recognize the resources the customer has brought to firm its consumption. That is how we expand grid deliverability rather than simply divide a fixed amount of infrastructure among more users.

The principle is simple: the grid provides shared infrastructure; private actors underwrite more of their incremental risk; dispatchability and private firming make that underwriting credible; and good market design recognizes the performance.

Why Speed to Power Is a Risk-Allocation Challenge and How Flexible Interconnection Can Accelerate Development

Steven: Why has Speed to Power become such a central issue for AI infrastructure, and what role can flexible interconnection play in helping projects come online faster?

Arushi: Speed to Power is fundamentally a risk-allocation problem.

Traditional interconnection assumes that before a large customer consumes power, the grid should be able to plan and build enough infrastructure to make that consumption firm. That is an extraordinarily high standard when loads are arriving faster than major transmission infrastructure can be planned, permitted, and constructed. It is like you made a well for a village to drink from, and one day an entire army comes for respite and relief, and expects this social infrastructure to merely perform bigger, harder, and faster to accommodate everyone instantaneously. That does not sound right, does it? 

The grid is a social good.

So, how do we maximize private investor returns, and participate economically and fairly with this social structure to inculcate prosperity for all? The clear alternative is to separate what the grid can guarantee from what the private customer is willing to underwrite.

That is the architecture I built with given market fundamentals, design strategy, and advocacy work into ERCOT’s Provisional Controllable Load Resource, or PCLR, framework. The planning system can preserve a conservative firm floor while allowing the customer to consume above that floor as a dispatchable resource, subject to actual system conditions. The customer gets an opportunity to energize sooner; the grid does not have to pretend every megawatt is firm; and the private sector has a powerful incentive to firm the gap with storage, generation, computational response, software, contracts, or some combination of those technologies.

That is what flexible interconnection should accomplish. It should not weaken reliability standards. It should create a bankable mechanism for determining which risks the shared system must carry and which risks a sophisticated private customer can carry itself.

Speed to Power becomes possible when we stop treating “firm on Day One” as the only legitimate form of grid access.

How Grid Flexibility Is Changing Site Selection, Financing, and Competitive Strategy

Steven: For developers and investors, how should grid flexibility factor into siting, financing, and competitive strategy as AI infrastructure scales?

Arushi: It should be part of project underwriting from the beginning.

A developer should not simply ask, “How many megawatts are available at this site?” The better questions are: How many megawatts are actually required to be firm? Can we find the right load profiles to firm + dispatch? Can we pair up with neighboring battery projects before we undertake an interconnection request for the load, and send that dispatchability information in with the load study? What constraints could interrupt that consumption? What technologies or contracts can firm the difference? And can the grid operator recognize and rely on those solutions?

That changes siting, which is the key. If those questions are being asked, more power service comes to loads that intentionally site infrastructure where it will serve more people and more industry. Two sites with the same nominal power availability may have very different deliverability, curtailment exposure, and opportunities for private firming, depending on the sites’ ability to mitigate thermal constraints identified in nodal studies.

It changes financing as well. If the grid provides a conservative floor and the customer wants substantially more power, the gap between those numbers is an underwriting problem. Investors should understand the probability and duration of dispatch to resolve power deliverability in peak conditions, the cost of mitigating that exposure, and the enforceability and performance of whatever batteries, generation, controls, or compute flexibility are being used to reduce it.

And increasingly, that becomes competitive strategy. The strongest AI infrastructure companies will not simply be the companies that secure the most power. They will be the companies that become exceptionally good at firming power.

That capability can become a moat: physical firming through batteries and generation; market and control architecture that properly attributes resources to the load; and internal computation that can move demand across time or geography in response to grid conditions. I call that Bring Your Own Everything. The more reliably a customer can underwrite its own incremental grid-access risk, the more credible its Speed to Power strategy becomes.

What ERCOT’s Market Reform Can Teach FERC-Regulated Markets About Deliverability, Dispatchability, and Growth

Steven: You have helped pioneer recent market structure reforms in ERCOT so that they keep pace with demand growth. What are some of your “lessons learned” from that experience, and what guidance should stakeholders take from your experience in ERCOT to inform their approach to FERC-regulated markets?

Arushi: My biggest lesson from ERCOT is that successful reform starts with the physical operating problem, not with a giant theoretical rewrite of the market.

ERCOT already had an important operational building block: large controllable loads could become visible, nodal resources whose response could be enforced through the market. The missing link was recognizing that operational dispatchability in the planning and interconnection process. My original PGRR134 proposal grew out of a very simple question: if ERCOT can reliably dispatch a load when the transmission system is constrained, why can’t that capability have value in getting the load energized sooner? It would be essentially the same SCED-enabled architecture that drives Connect and Manage for generators. It is the reason generators already get Speed to Power in ERCOT. 

The answer became a framework in which the customer effectively signs on the dotted line — Form W in the new Batch Zero load policy documents. Dispatchability is not a promise or a press release. It is an enforceable operating obligation exchanged for a valuable grid-access benefit. I filed the original planning-guide proposal, built a coalition of more than 35 market participants around the concept, and ultimately saw that architecture incorporated into ERCOT’s Batch Zero framework. I was glad for Mintz’s support, and that of its clients as co-advocacy champions. 

There are several lessons I would carry into FERC-regulated markets.

First, preserve planning rigor. You do not need to make firm transmission planning less conservative in order to create a conditional or non-firm pathway.

Second, make dispatchability real. If the system operator cannot send an instruction and receive a specific, measurable, and enforceable response from the load, then “flexibility” is mostly a slogan.

Third, separate the firm floor from the operational opportunity above it. That gives the planner confidence, the operator controllability, and the customer a pathway to earlier energization. Doing that means also ungating metering, telemetry, SCED, or SCUC architecture so that private players can build for dispatchability. Otherwise everything will remain sticky. 

Fourth, credit private firming when it can actually be proven. If a customer brings storage, generation, computational response, or a contracted resource that reliably reduces the grid’s exposure, market rules and planning studies should be capable of recognizing that contribution rather than studying the load as though the solution does not exist.

And finally, build the reform around a product people can actually finance. The grid operator needs confidence in performance, the utility needs a workable operating arrangement, the customer needs to understand its curtailment exposure, and investors need to understand who carries which risk.

That is the deeper ERCOT lesson for FERC-regulated markets: do not ask the grid to underwrite every private return. Build mechanisms that let private actors underwrite more of their own access while making the shared infrastructure more deliverable for everyone.

Explore more insights in our Inside Digital Infrastructure Q&A series.

About the Author

Arushi Sharma Frank

Arushi Sharma Frank, Esq., is an expert in energy market economics, law, regulation, and technology. Her work spans the intersection of policy, markets, and infrastructure, shaping how complex energy systems evolve under changing regulatory, economic, and technological conditions. She has served in strategy execution and policy leadership roles at the American Gas Association, Electric Power Supply Association, Exelon Constellation, and Tesla. She now runs Luminary Strategies, a consulting practice focused on high-impact work at the frontiers of power system transformation. She advises private equity teams and portfolio companies engaged in utility grid technology, retail energy, battery storage, hyperscale development, and load flexibility solutions. The firm also supports emerging nonprofit and advocacy organizations advancing American energy dominance, competitive energy markets, and industrial policy objectives. Her areas of expertise span all facets of her work driving energy policy reform and business innovation. They include behind-the-meter grid integration, commercial energy project execution and operations, energy and capacity market design, large-load interconnection with co-located generation and storage, and transmission and distribution interconnection and utility grid planning. She has also led work on complex electric infrastructure development, natural gas commercial regulation, energy metering and telemetry requirements, and the formation, launch, and compliance of both wholesale and retail energy companies.

Subscribe To Viewpoints