Gigaton, the London-based AI startup formerly known as Carbon Re, has raised a $26 million Series A to take on one of industry’s perhaps least glamorous but most valuable software markets: the control systems running cement, steel, glass and chemicals plants.
The round was led by Plural, with participation from 2150, Semapa Next, Planet A Ventures, Cambridge Enterprise Ventures, the UCL Technology Fund and Clean Growth Fund. The raise brings Gigaton’s total funding to more than $35 million.
The new capital will be used to grow the team fivefold and expand beyond cement into steel, glass and chemicals.
Gigaton began life as ‘Carbon Re’, which was about AI for industrial decarbonisation. But CEO Josh Vernon told Pathfounders the company learned that simply putting AI “on top of existing control systems” was “theoretically correct, but ultimately naive.” The company now wants to go deeper into the plant itself.
Gigaton’s thesis is that the current industrial control stack was not built for today’s volatility. Cement plants that once burned coal are now being asked to use a wide variety of ‘fuels’ including recycled waste, shredded tyres and even sewage. That may be cheaper and lower-carbon, but it also makes the production process far harder to control.
As Vernon puts it, “You cannot build that future on heuristic-based or rules-based logic that was written by people in the 1980s.”
Gigaton’s system runs models at the edge, connects directly into the plant, simulates process behaviour, and then autonomously adjusts variables such as fuel mix, kiln speed, fan speed and oxygen levels. Vernon describes the job as “control in the world’s largest robots.”
The company is starting with cement, which Vernon says is the most-used product on Earth after water and accounts for around 8% of global CO₂ emissions. The software platform is already deployed with producers including Adani Cement, Heidelberg Materials, Holcim and Mannok. Gigaton claims its systems can deliver $1 million to $3 million in annual operational savings per plant, alongside emissions reductions of up to 30,000 tonnes of CO₂ per site.
That puts it in a tough, incumbent-heavy market. ABB has Expert Optimizer, FLSmidth has ECS/ProcessExpert, Rockwell has Pavilion8, while AspenTech and Honeywell also offer advanced process control platforms. Vernon was careful not to dismiss them, saying they “built the industrial world” and still control much of the hardware inside plants. But he argues their software is often closer to “if this then that” logic than modern AI, which has huge limitations.
Gigaton’s bigger ambition is the fully autonomous “dark plant”, or industrial sites that can run with minimal or no on-site operators. Vernon says China is already building fully autonomous "dark plants," while Western industry is still moving through the intermediate stage of remote-controlled operations.
The company’s pitch, then, begins to resemble arguments about industrial sovereignty, and will appeal to those who say Europe and the West need to accelerate AI-native control systems for heavy industry if they are to keep up with China.
