Mantic Raises $25 Million After AI Forecasting Systems Beat Human Competitors

Mantic raised a $25 million seed round after AI systems outperformed human competitors in a forecasting tournament, pushing probabilistic prediction into a new phase.

Forecasting has long been treated as a human judgment problem: gather evidence, weigh uncertainty and assign a probability to what happens next.

A new generation of AI systems is challenging that assumption. Reuters reported on September 18 that London-based Mantic raised $25 million in seed funding after AI entrants performed strongly in the 2026 Metaculus Cup, with Mantic beating human competitors in forecasting political, economic and cultural developments.

The interesting shift is not that AI can make predictions. It is that AI systems are beginning to compete on calibrated probabilities — a skill with direct value in business decision-making.

Forecasting is different from answering

A chatbot can produce a confident paragraph without knowing how uncertain its answer should be. Forecasting requires a system to express uncertainty numerically and update its belief when new information arrives.

That makes evaluation clearer. If a system repeatedly assigns high probabilities to events that do not occur, its calibration can be measured.

Why companies may care

Businesses constantly make probabilistic decisions: demand planning, hiring, product launches, supply chains, market expansion and risk management.

An AI forecasting system could help teams aggregate large amounts of information and continuously update scenarios. It would not remove uncertainty, but it could make assumptions more explicit.

The strongest use case may be decision support rather than automated decision-making.

Executives can compare the model’s probability with human forecasts, inspect the evidence and investigate where views diverge.

Funding follows measurable performance

Radical Ventures led Mantic’s round, with participation from Microsoft’s M12 and other investors, according to Reuters.

The funding shows why objective competitions matter for AI startups. In a crowded market, a measurable performance result can provide a stronger signal than a generic claim that a model is “smarter.”

What to watch

Forecasting systems will need to prove that tournament performance transfers into messy real-world environments where data is incomplete and incentives matter.

They will also need strong provenance so users can understand which information influenced a forecast.

If AI can become consistently well-calibrated about uncertain future events, forecasting could emerge as one of the most valuable analytical applications of advanced models.