Energy has become an all-consuming concern for data center developers and operators. As power availability becomes increasingly scarce and grid dynamics become more volatile, energy planning has emerged as a high-stakes discipline. Mistakes aren’t just expensive; they can delay or even derail data center projects.
To gain a clearer understanding of how the industry addresses these challenges, I spoke with Holt Bradshaw, principal consultant at Siemens PTI Energy Advisory. Bradshaw works at the intersection of infrastructure planning and power systems, giving him a front-row seat into the evolving energy strategies of some of the world’s largest data center operators.
Let’s start with the current landscape. What are the biggest challenges data centers face today when it comes to securing adequate, reliable power?
Bradshaw: They range from equipment and power procurement hurdles to delays in interconnecting to the transmission grid.
There are currently very long lead times to secure power. It takes up to five years to obtain a large gas turbine for thermal generation. Transmission expansion, which may be needed to connect new projects to the grid, may take up to 10 years.
Siting is also an issue. Prime sites for data center projects are becoming increasingly scarce. The best sites offer access to inexpensive and reliable power, high connectivity and low latency for data traffic, access to water for cooling, and minimal exposure to natural disasters. Such sites tend to be crowded with data centers or proposed data centers. This creates a long waitlist of small projects seeking to interconnect with the distribution grid and large projects awaiting interconnection to the transmission grid.
“The queue to interconnect extends for multiple years in some places.” — Bradshaw
The queue to interconnect extends for multiple years in some places. This results in some data center developers seeking to develop their own on-site power supply solutions, often termed Bring Your Own Power, or BYOP. But on-site solutions can come with their own challenges, ranging from significant investments to reach desired power reliability levels and cope with fast oscillations in power requirements to the long lead times for equipment delivery.
What are the highest priorities for data centers in energy site selection?
Bradshaw: Each data center developer considers power supply in terms of power reliability, sustainability, affordability and speed.
Energy reliability is essential. Data centers with the highest reliability requirements (Tier IV) generally pay the highest leasing fees, which, interestingly, are often priced in dollars per kilowatt per month.
Many data center developers prioritize sustainable power supply, though given the substantial power requirements, acquiring sufficient low or no carbon-emitting power can be challenging.
Of course, affordability is also important. Power costs are a substantial portion of the total data center operating cost and cannot be ignored.
Another high priority is speed-to-power – how quickly data centers can secure power so that they can begin operating. That’s because tech is a winner-takes-all industry; the first mover has a disproportionate advantage, especially when it comes to AI.
Unfortunately, it’s neither cheap nor easy to achieve speed-to-power at present, resulting in an incongruity in development timelines. Data centers can be built quickly; however, power plants and transmission lines cannot.
Are there common energy planning mistakes that data centers make?
Bradshaw: Yes, and they are understandable because data center developers and operators are generally not experienced in the complexities of the electric grid operations and power markets.
“Data center developers often do not understand the time and technical requirements of connecting to the grid.” — Bradshaw
The first error is misaligned expectations. Data center developers often do not understand the time and technical requirements of connecting to the grid. To be fair, the interconnection requirements vary by utility and Independent System Operators (ISO)/Regional Transmission Organizations (RTO). And they are in flux. The requirements often fail to recognize the U.S. is facing demand growth not seen in decades. Many utilities and ISO/RTOs, designed to serve flat or in some cases declining load, are overwhelmed with both load and generation interconnection requests, all while the requirements are changing in real time.
As a result, some data center developers are considering building their own power supply. However, they may not understand that choosing generation equipment is only part of ensuring energy reliability; there are other elements, ranging from reliability of their own internal grids to access to dependable fuel sources, depending on the on-site solution.
Do you see them making errors that can lead to inaccurate cost projections?
Bradshaw: Yes, their energy plans sometimes do not fully account for cost and schedule complexities.
As I said earlier, data center developers need to understand the tradeoffs between power reliability, sustainability, affordability, and development speed – you cannot achieve each of these at the same time. Extremely reliable power supply requires significant redundancy. Sufficient sustainable power cannot be delivered without extensive contracts or the addition of storage. With the 5+ year backlogs at major combustion turbine vendors, smaller, more modular options that cost more are now in play. The need for speed sometimes drives data center developers to make choices that address short-term needs without considering the long-term financial costs and benefits.
Here are some examples.
- The batteries often required to support renewable sources of power or to support the grid in emergencies are capital-intensive and, during their lifetime, often require refurbishment to maintain capacity. Data centers sometimes overlook this consideration. They may not budget the cost of maintenance interventions needed later in the asset’s operating life.
- Issues also arise in monetizing on-site energy assets. A data center may choose to rely on on-site energy until it can secure a grid connection. But how will it monetize the on-site asset after the data center connects to the grid? It’s important to know the answer to this question before choosing and installing an on-site energy asset. Will you be exporting surplus generation? Sell other grid services? Knowing these answers and having a monetization plan will ensure the correct selection of assets from the start.
Are there market uncertainties data centers should have on their radar now?
Bradshaw: There are both short- and long-term policy uncertainties.
Perhaps the most significant short-term issue is the time required to complete the large interconnection process. These processes are somewhat opaque and vary by utility and market. The technical analysis requirements and standards a new load must meet in order to connect have caused delays and uncertainty, though these are beginning to become clearer in some places.
Another key policy concern revolves around the use and costs of imported raw materials, equipment, etc. In order to build generation fast enough to meet rapidly expending load in the timeframes expected, we will need access to the complete global supply chain, hopefully at manageable costs. Recent tariff developments not only increase cost, but they also realign supply chains, causing disruption and delays.
Wholesale regional market prices can also have an impact on large energy users . These markets are complex with sensitive and volatile pricing. PJM, the largest grid operator in the US, held a capacity auction that resulted in a 22% price increase from a year earlier. Data centers in that footprint might be impacted more or less severely, depending on whether they procure their energy in the wholesale market or via an intermediary, and on the hedging arrangements they have in place. It’s imperative that data centers understand how to procure power and fuel, and hedge against this kind of market risk.
Whether they use on-site gas generation or purchase from the market, they also need to manage risks associated with fuel price volatility.
Q: What alternatives can data centers pursue if they face long interconnection delays?
Bradshaw: Some are turning to on-site power, with microgrids that include both generation and energy storage. The costs of this solution can increase fast with the required reliability and with the flexibility needed to follow rapid changes in power load. On-site power may be a temporary solution employed until the data center can establish a grid connection. Or it can serve as permanent off-grid energy supply.
To secure the strongest energy reliability at the lowest price, it often makes sense to combine the two — connect to the grid and install on-site power.
However, on-site power also presents challenges.
First, local environmental regulations may prevent the data center from using the generation technology of its choice.
Second, the data center may face slow delivery of certain types of generation and energy storage equipment because of the supply chain constraints I described earlier.
When used as the sole source of power, on-site generation can be more expensive than grid power.
What other considerations should data centers take into account?
Bradshaw: We’ve discussed energy reliability and cost. Sustainability is a third concern.
Some tech companies have internal emissions requirements to meet. There are various approaches they can take, ranging from power purchase agreements with renewable generation, purchase of carbon offsets or on-site renewable energy. Again, risk management comes into play.
In making sustainability decisions, it’s crucial to analyze synergies that can improve not only the project’s emissions profile but also its cost and reliability.
Community opposition is another risk faced by data centers now. What can they take to gain local acceptance?
“Large loads coming onto the grid may raise rates in some locations but decrease them in others.” — Bradshaw
Bradshaw: Some of the opposition centers on power prices — the idea that data centers will drive up electricity rates. However, it’s essential to note that the impact of data centers on rates may change depending on the location and the specific type of data center. Large loads coming onto the grid may raise rates in some locations but decrease them in others. And data centers with more flexible applications, which can shift some of the power consumption in time in response to prices reflecting grid stress, can impose a less significant burden on the grid.
Where data centers raise rates, one solution is to charge them a separate tariff. We’re seeing some of our data center clients participating in newly created data center-specific tariffs, which often require increased upfront costs and long-term commitments to minimize or eliminate ratepayer impacts.
Any last words of advice for data centers facing the risks you described here?
Bradshaw: Determining the best supply solution for each data center requires in-depth knowledge of technology, operations, pricing and markets. Whether you have an in-house team or external support, it’s crucial for data centers to work with someone who speaks the language of utilities, understands technical problems associated with power assets, and can navigate both regulated and deregulated power markets.
A knowledgeable partner can help you secure a grid connection, screen sites, access financing, choose equipment, interact with utilities and grid operators, find the right balance between on-site and grid-produced power and monetize and optimize your system.
It’s about building a bridge between your operation and the energy world so that you secure energy with the optimal balance between reliability, cost and sustainability.
This article was written in partnership with Siemens as part of the Energy Changemakers Ambassadors program.
