British AI startup Mantic raises $25M to build ‘superhuman’ AI forecasting after Metaculus win

British AI startup Mantic has raised $25 million in seed funding after its forecasting system beat every human competitor in the summer 2026 Metaculus Cup, giving the young company a striking proof point for its claim that machines are getting better at predicting uncertain real-world events.

Radical Ventures led the round at an undisclosed valuation, joined by Microsoft’s M12 venture fund, Thinking Machines Lab, Balderton Capital and other investors. Radical Ventures partner Aaron Rosenberg has joined Mantic’s board. Sky News first reported the financing, with Reuters later confirming the $25 million round and additional details.

The London-based startup is tackling a branch of AI that gets far less attention than chatbots or coding agents: judgmental forecasting.

Rather than predicting weather or running conventional statistical models against clean datasets, Mantic focuses on messy questions involving business, technology, geopolitics, policy and culture, often weeks or months into the future. The company says its systems are built for situations where research, context and judgment matter as much as historical data.

Mantic co-founder and CEO Toby Shevlane described the goal in ambitious terms.

“We’re now upgrading the level at which humans can understand the future.”

Shevlane founded Mantic with Ben Day in 2024 after working as a research scientist at Google DeepMind. His experience there helped spark the idea after he found himself needing better forecasts for global developments that could affect AI.

From prediction contest to commercial product

Mantic does not build its own frontier foundation model from scratch. Instead, the startup adapts models developed by other AI labs for forecasting tasks. Its system is tested against historical events, scored on its predictions and refined based on those results.

That approach received a major public test in the latest Metaculus Cup, an online tournament where participants assigned probabilities to political, economic and cultural outcomes.

Mantic finished ahead of every human contestant and behind just one other AI system, a bot named laertes. The broader showing from AI systems marked the first time machines dominated the competition, according to Reuters.

One example exposed a weakness that has long troubled human forecasting: following the crowd.

Participants were asked whether Shakira’s “Dai Dai” would surpass “Waka Waka” on the Billboard Hot 100. Human forecasters heavily favored the consensus view that it would not. That call proved wrong. Mantic was less committed to the consensus and avoided taking the same large miss.

The system showed a similar willingness to break from the crowd during Colombia’s presidential race. Mantic assigned Abelardo De La Espriella roughly a 40% probability of winning when the consensus estimate stood near 30%. De La Espriella later won the election.

One tournament cannot establish that an AI system can reliably foresee every type of future event. Forecasting performance can vary by question, time horizon, and available information. Mantic’s result does show that AI systems are becoming credible competitors in a field once dominated by highly skilled human forecasters.

That has obvious commercial appeal.

Rosenberg said companies and government agencies worldwide have shown interest in Mantic, with some already using its technology. The startup has not disclosed customer names. Hedge funds and trading firms have shown particular interest, where a small forecasting advantage can translate directly into financial returns.

“If Mantic is superhuman as it is, they can make money off of that immediately,” Rosenberg said.

That may be the bigger story behind the $25 million round. AI has spent the past few years learning how to generate text, images, and software. Mantic is betting that the next valuable capability is something far harder to fake: being right about what happens next.