What will power the grid in 2035? The race is wide open

AI’s insatiable demand for power has tech companies hunting for new energy sources — a search that has fueled competition and investment into fusion and fission startups.

For many, natural gas is the easy answer for 24/7, baseload power. It’s tested, inexpensive, and widely available. But the war in the Middle East exposed its vulnerable supply chain after Iranian drone strikes took out a significant portion of natural gas infrastructure in Qatar, a major exporter. At the same time, surging demand has created a waitlist for gas turbines so long that today’s orders probably won’t be fulfilled until the early 2030s.

Those delays not only pose a risk to tech companies, but also to the natural gas industry itself. 

In the U.S., 40% of the natural gas consumed today goes toward generating electricity. By the time the turbine shortages relent, the industry could be flush with a fresh crop of competitors. Both small modular nuclear reactor (SMR) startups and fusion power startups plan to start connecting their first commercial power plants to the grid in the next five to seven years, about how long it takes to get parts for a new natural gas power plant.

Nuclear threat

SMR startups might have the best shot at displacing natural gas power plants. In many instances, the technology tweaks the designs of existing fission reactors, but the fundamental physics has been proven and widely used for decades.

Several SMR companies aim to have reactors up and running before the decade is over. Kairos Power, which counts Google as a future customer, is one of them. The company received approval for its Hermes 2 demonstration reactor in 2024, and construction is well underway. Oklo, which merged with Sam Altman’s blank check company in 2024, is targeting 2028 for its first commercial operations, according to its annual report.

Others hope to follow a few years later. X-energy, which counts Amazon as an investor, is aiming for the early 2030s, while the Bill Gates-founded TerraPower, which has a deal with Meta, is planning to begin commercial operations in 2030.

To displace natural gas as the generating source of choice, SMRs will need to scale quickly, realizing the economies of scale that their business models depend on. That won’t be easy. But tech companies appear confident enough that they’re either investing in startups or signing agreements with them for gigawatts worth of power. 

Fusion’s timeline

The other technology companies are warming to is fusion power. Though it isn’t as proven as fission, nuclear fusion promises to deliver large amounts of power using little more than seawater as fuel. 

Fusion startups are also targeting the early 2030s — or sooner — to deploy their first reactors. Fusion power 

One front-runner, Commonwealth Fusion Systems, is on track to flip the switch on its demonstration reactor next year. Its first commercial reactor, the 400-megawatt Arc, is expected to start generating power in Virginia in the early 2030s.

Another startup, a relative newcomer, hopes to start construction on a grid-scale power plant in 2030. Inertia Enterprises has based its technology on the reactor design employed by the National Ignition Facility, which was the first to prove that controlled nuclear fusion reactions could generate more power than they consume.

But Helion may have the most aggressive timeline out of all of them. The Sam Altman-backed startup is racing to build Orion, its first commercial-scale power plant, by 2028 to supply Microsoft with electricity. The company is also reportedly in talks with OpenAI to provide up to 5 gigawatts by 2030 and 50 gigawatts by 2035. To hit those numbers, Helion will have to build 800 reactors by the end of the decade and another 7,200 in the five years after that. 

If the startup can deliver power in those quantities, it would completely rewrite the energy market. Last year, the U.S. added 63 gigawatts of new generating capacity across all sources. If Helion can build close to 10 gigawatts of new capacity every year, the company alone would add more power than the entire natural gas industry did last year.

The price problem

The challenge for all those companies — including gas turbine manufacturers — is cost. 

SMR startups are counting on mass manufacturing to drive cost reductions, but that hypothesis has yet to be proven. Today, nuclear power is one of the most expensive forms of new generating capacity at around $170 per megawatt-hour, according to Lazard. Fusion faces a similar scale-up challenge, though it faces even more unknowns. Some experts predict one megawatt-hour from a fusion power plant could run about $150 initially. 

New baseload natural gas power plants, meanwhile, run about $107 per megawatt-hour, per Lazard, though prices have been trending up in recent years, perhaps setting it on a collision course with both new fission and fusion reactors.

But they might all be undercut by renewables paired with batteries. 

The costs of wind and solar power have dropped precipitously over the last decade. Wind power appears to have hit a bit of a plateau in recent years, but solar prices continue to inch downward with no signs of stopping. Batteries, too, have grown cheaper over the years, to the point where grids are installing massive quantities of them — 58 gigawatts-hours last year. Even without subsidies, solar paired with batteries ranges from $50 to $130 per megawatt-hour, overlapping fusion, fission, and natural gas. 

Those figures are all with current battery technology derived from chemistries intended for electric vehicles. Newer designs aimed squarely at grid connections could slash prices further. Form Energy, for example, recently signed a deal to provide Google with electricity from a 30 gigawatt-hour iron-air battery. Another, XL Batteries, can repurpose old oil tanks to store its inexpensive organic fluid — the size of the battery is only limited by the size and number of the tanks.

Because those new batteries eschew the use of critical minerals like lithium, cobalt, or nickel, they promise to dramatically reduce the cost of long-duration energy storage to the point where it’s hard to make a case for anything else.