Parallax Builds 10 MW Gas Turbines for the AI Power Gap

Parallax is building an approximately 10 MW gas turbine for onsite AI data-center power. The California company was founded in 2026 and is led by founder and CEO Carl Schoeller, who previously co-founded the GPS-denied drone-navigation company Theseus.

Parallax matters because the AI infrastructure market is discovering that compute procurement and power procurement run on different clocks. A developer can secure chips and construction capital while waiting years for grid connections or generation equipment. Parallax is designing a smaller turbine around manufacturing speed, common materials, modular transport, and earlier delivery, even when that choice gives up some fuel efficiency.

The company emerged from stealth on September 24, 2026 with a $117M fundraise backed by Founders Fund, Eclipse, Lux Capital, Diffusion, Greylock, and General Catalyst. The money gives Parallax a serious industrial budget. It does not yet give the company a proven machine, customer fleet, or operating record.

About Parallax

Parallax describes itself as “The AI Energy Company.” Its first product is a gas turbine sized at approximately 10 MW, small enough to move in shipping containers and combine in multiples for data-center sites with much larger power requirements.

That scale positions Parallax between two awkward alternatives. Large turbines can deliver substantial output but face long production schedules and specialized supply chains. Smaller temporary generation can arrive faster but may not fit the cost, durability, or operating requirements of a permanent AI campus. Parallax is trying to turn a transportable machine into repeatable blocks of onsite capacity.

The company has not disclosed customers, orders, revenue, a production facility, or commercial terms. The launch is therefore best read as a product thesis with financing behind it, not a completed operating business. Parallax is targeting a first prototype by the end of 2026, testing in 2027, and customer deliveries in 2028.

How Parallax Is Redesigning Turbine Manufacturing

Conventional high-performance gas turbines use specialized alloys, extreme combustion temperatures, and cores assembled from more than 2,000 parts. Those choices support efficiency and durability, but they also create manufacturing complexity and dependence on narrow supplier networks.

Parallax is deliberately changing the optimization target. The company says 3-D printing can reduce the core to a handful of parts. Lower combustion temperatures permit more common materials and a broader aerospace supply base. Carl Schoeller also says the selected materials have a global supply five orders of magnitude greater than current turbine-blade materials and that the design reduces core part count by three orders of magnitude. Those two figures are company claims and have not been independently validated.

The engineering trade is explicit. Lower temperatures can reduce fuel efficiency, increasing operating cost over the machine’s life. Parallax is betting that an operator may accept that penalty when the alternative is waiting years for generation equipment while a data-center investment sits idle.

Why the AI Power Market Is Pulling Hardware Forward

The International Energy Agency estimates global data-center electricity consumption could rise from 485 TWh in 2025 to about 950 TWh in 2030. The agency also estimates that 15 GW to 27 GW of onsite natural-gas generation could serve data centers by 2030, mostly in the United States.

U.S. forecasts remain wide because chip shipments, utilization, cooling systems, and facility timelines can change quickly. The Department of Energy and Lawrence Berkeley National Laboratory estimated that American data centers could consume 325 TWh to 580 TWh in 2028, equal to 6.7% to 12% of national electricity consumption.

Those numbers do not guarantee demand for Parallax. They do explain why developers and investors are testing multiple ways to compress the power schedule, including DevCuration’s coverage of TAR’s off-grid AI power strategy and Bluecore Energy’s floating nuclear plan. The market is not converging on one fuel or machine. It is converging on the cost of delay.

Carl Schoeller and the Industrial Founder Test

Carl Schoeller previously co-founded Theseus, a defense-technology company developing GPS-denied drone navigation. TechCrunch reported that Theseus grew from a 3-D-printed hackathon drone and raised a $4.3M Seed round in 2025. That background connects software, autonomy, rapid prototyping, and physical systems, but a gas turbine introduces a different level of thermal, mechanical, regulatory, and service risk.

Parallax has not publicly identified a current CTO, broader executive roster, or additional founder. The company also does not maintain a public careers page or open-role list. Its official site invites builders to contact build@parallax.ai, which signals active team formation without revealing headcount, functions, location requirements, or hiring volume.

The investor group adds capital and industrial credibility, but it cannot substitute for the operating team Parallax will need. Turbine development crosses materials science, additive manufacturing, combustion, controls, power systems, permitting, field service, and supply-chain execution. Hiring across those disciplines will be a meaningful signal only when the company discloses the roles and the team responsible for the machine.

What Parallax Still Has to Prove

Parallax’s technical proof starts with sustained output, thermal behavior, manufacturing repeatability, component life, reliability, maintenance intervals, and safe operation. Commercial proof requires customer orders, a production system, permits, fuel supply, emissions compliance, site integration, service capacity, and economics that justify lower efficiency.

None of those outcomes is public yet. The company has not disclosed independent performance results, emissions data, unit cost, lifetime fuel cost, installed capacity, or contracted deliveries. Its 2026, 2027, and 2028 milestones are development targets.

That uncertainty is normal for a company emerging from stealth, but the hardware does not grade on a startup curve. A machine connected to a data center has to work through heat, vibration, maintenance, changing loads, and regulatory review while the customer measures every lost hour.

What Parallax Signals for AI Infrastructure

Parallax represents a broader capital shift from compute alone toward the physical systems that determine when compute can operate. Chips remain scarce and expensive, but the value of a chip changes quickly when power delivery moves beyond the facility schedule.

The company’s proposition is therefore about time as much as energy. By simplifying the core, broadening the material base, and shipping smaller units, Parallax wants to make generation equipment behave more like repeatable infrastructure and less like a bespoke industrial monument.

The $117M fundraise gives Parallax room to test that proposition at full scale. The durable company will be built only if manufacturing speed survives the test stand, the first installation, and years of maintenance. For AI developers facing a power date, that evidence will matter more than the elegance of the launch.