How fuel cells can solve the AI data center power problem

We talk a lot about the AI data center power problem - and by this point, I’m sure you’re familiar with what’s happening. AI data centers need a lot of power. We need a lot more power to serve them. Electricity grids cannot keep up. Data centers are rapidly shifting to onsite power solutions. But there are few solutions. Gas turbines have 5-year waiting lists - and big emissions. SMRs have long lead times. Renewables are intermittent - so have complex energy storage requirements. And so on.

Within this context, I get one question a lot – what do we do?

So today, I’m deep diving on fuel cells, one of the few solutions available now that offer clean, baseload, dispatchable onsite power. The most obvious example of this is Bloom Energy, whose stock price has been on an absolute tear the last 18 months. As I write this, Bloom is currently valued at $82.5B and trading at more than 40x revenue. It was a long journey for Bloom to get to this point – and the challenges of costs, complexity and capacity are not fully solved - but it’s clear from this growth that the time for fuel cells is now. It’s also clear that there’s a massive market opportunity for other fuel cell companies to emerge, particularly those that can solve fuel cells’ historic cost, complexity and capacity challenges with new technology IP. But before we get into that, let’s start with a simple question – what exactly is a fuel cell?

Fuel cell technology (stacks) converts fuel to electricity using an electrochemical process, instead of combustion. Fuel cells generate electricity as long as fuel is put into them, naturally making them dispatchable baseload power. Since there is no combustion, fuel cells are also naturally a clean energy technology - they release no air pollutants and significantly fewer carbon emissions than turbines or other gas-based technologies. This streamlines permitting - and in the US, makes them eligible for pretty significant tax credits. When you add carbon capture technology to an integrated fuel cell energy system, they are of course net zero. If you believe that a large portion of our global economy will continue to run on gas (I do), fuel cells should be part of any conversation about energy transition.

Fuel cells, however, can also run with any fuel source - gas, biogas, ammonia or hydrogen. The internal electrochemical reaction inherently comes from hydrogen, however any fuel source can be refined into hydrogen - either internally or externally to the fuel cell technology itself. This means that fuel cell energy systems are fuel agnostic - offering significant optionality to customers, and the opportunity to transition from gas over time as other fuel supply chains develop. This is another reason that fuel cells should be a part of any conversation about energy transition.

Fuel cell stacks are also inherently scalable in form factor. This means that they are applicable to a wide range of customers and use cases, from small mobile applications (eg. cars, trucks, mining equipment, drones, ships, planes) to large stationary applications (eg. data centers, airports, refineries, utilities, manufacturing). This makes the market very large and the energy transition conversation even more poignant.

When a fuel cell generates electricity, the electrochemical reaction naturally produces heat as a byproduct. Instead of letting this heat escape and go to waste, a heat recovery system can also be added to a fuel cell power system (called a combined heat and power system or CHP). This can capture and redirect the heat to be used for space heating, water heating or other industrial processes, providing further value to customers. Because the fuel is being used twice - once for power and once for heat - it requires less total fuel to meet a building’s energy needs, subsequently reducing the carbon footprint - and furthering the energy transition impact.

Finally, fuel cells should also be a part of any conversation on behind-the-meter or onsite power. The nature of a fuel cell is that it’s hardware - manufactured, deployed to a site and inherently modular and scalable. Most industrial fuel cell companies have modules that generate under 1MW – and are of course scalable up to GW scale designs. This makes them a great option for both on-grid and off-grid power of different scales — and a key element of the accelerating market demand for onsite microgrid energy solutions. As such, they are perfectly suited to serve the growing data center demand for onsite power.

Fuel cell technology, however, is not new. It’s been around for a long time. The first usable fuel cell was developed in 1932 by Francis T. Bacon, and then later refined in the 1960s by NASA which utilized fuel cells to provide electricity, heat and drinking water during the Gemini and Apollo space missions. Over the last 30 years, fuel cell technology has transitioned from aerospace to broader commercial and industrial applications.

Despite all the positives I discussed in the prior section, fuel cells have historically been held back by high costs and capacity. Producing fuel cells has historically required highly capital-intensive specialist processes to manufacture the precise, microscopic components in the fuel cell stack. This has translated to high capital purchase costs and high long-term energy costs (LCOE), which have inhibited widespread fuel cell commercialization and customer adoption. This has also translated into expensive manufacturing facilities with low production volumes and challenges meeting the demands of commercial scale.

Anyone who has followed the fuel cell space, mentions to me that nearly every fuel cell company has struggled over the last 20 years. Fuel cell skeptics often like to point out that fuel cell technology is old technology, that nearly every fuel cell company has been (or nearly been) bankrupt, and that “it took Bloom 18 years” to get to where they are today. These things are all true. But they disregard two things. Firstly, regardless of how long it took Bloom or how difficult their journey has been, they are worth $83B and raking in massive data center, industrial and utility customers all over the world. Their pricing has also fallen materially in recent years, making the total cost of ownership for a data center competitive with many alternatives — and faster. Secondly, and perhaps more importantly, this view disregards the potential for entrepreneurs to bring significant new technology innovation to the market that can materially bring down prices, reduce complexity and improve manufacturing scalability and system durability - beyond what Bloom has achieved. Any company that can do that is poised for massive growth and adoption.

At SSV, we believe that we have entered the fuel cell era. Put simply, the time for fuel cells is now. The converging energy mega trends of onsite power, time-to-power and energy transition, accelerated by unprecedented power demand from AI data centers, make it a true heyday for fuel cell technology and systems. Bloom has proven this. But this is a massive market - and there’s room for lots of players, particularly those with innovations that can materially bring down costs and complexity. The global power market is expected to be more than $2 trillion by 2034. As a co-investor said to me recently: “We need many Blooms in this market.”

As we look ahead at SSV, we believe that further technology advancements and innovation in fuel cell energy systems will create numerous big new winners in the space. We believe that the nature of the current power supply and demand imbalances, and the inherent speed to market of fuel cells versus alternatives that often require some combination of regulatory adoption, permitting or energy storage integration, has the potential to drive exponential value creation for private market investors in the space going forward - just as we’ve seen in the public market with Bloom.

Onward,

Brynne

Managing Partner, Smart Society Ventures

🇨🇭 🌱☀️ Our friends at Voltiris secured a loan guarantee from the Swiss Technology Fund to scale their spectral filtering technology, turning greenhouses into dual-use infrastructure that transmits the light crops need while redirecting the rest to solar panels to generate onsite clean electricity without compromising harvests.

🇩🇪 ☢️⚡ Focused Energy raised $240M Series A to advance laser fusion power, using the capital to develop its technology and establish a major fusion deployment site at RWE’s former Biblis nuclear power plant in Germany

🇺🇸 ☢️⚡ Thea Energy raised $100M Series B led by the U.S. Innovative Technology Fund to scale stellarator fusion power plants, expand magnet manufacturing capacity and support construction of its Eos integrated stellarator system.