Zenithon raises 10 million dollar initial round led by Backed VC for physics AI models

Zenithon raises 10 million dollar initial round led by Backed VC for physics AI models

30 September 2026 — Zenithon, a London company building AI models for extreme-physics engineering problems, has announced a 10 million dollar initial funding round. The announcement was published on 30 September by participating investor Seraphim Space and reported by Tech.eu the same day.

Not yet independently verified. First source is a participating investor (Seraphim), not the lead (Backed VC) or the company. Stage is stated only as an initial round; one aggregator calls it seed, which is not used. Company website not confirmed from an announcement page; zenithon.com answered but its content was not checked. We will update this when it can be confirmed, and remove this note.

The round

Seraphim describes the round as a 10 million dollar initial funding round led by Backed VC. Other investors named are Seraphim, Lunar Ventures, MMC and SOSV, together with founders and directors from large cloud providers. Seraphim says the money will pay for model development, growing the team from 11 to 17 people, and commercial work in San Francisco and across the United States. The stage is given only as an initial round, so TrustList does not label it seed or Series A.

What the company does

Zenithon develops what it calls world models for extreme physics. The target uses named in the investor post include spacecraft design, advanced propulsion, fusion energy and semiconductor manufacturing. The stated problem is that conventional engineering simulations can take days or weeks, which limits how many design options a team can explore. The company says its approach aims to let engineers evaluate very large numbers of design configurations in near real time. It was founded by Alex Higginbottom, chief executive, and Abetharan Antony, chief technology officer. These are statements from the announcement, not tested performance claims.

Why it matters for buyers

Simulation software is a long-established part of engineering budgets, and AI surrogate models are being pitched as a faster layer on top of or instead of conventional solvers. A funded entrant in this space is relevant to engineering and R&D teams that spend heavily on compute for simulation. At this stage, though, the company is young, has a small team and describes its early focus as fusion energy, so general availability to other industries is likely some way off.

What to check

• Whether the product is available to customers now or is still in development, and for which sectors.
• How model accuracy is validated against established solvers, and what error bounds are published.
• Where customer design data is stored and how it is protected, which is especially important for aerospace and semiconductor work.
• Export-control and security constraints that may apply to advanced propulsion and similar applications.
• Pricing and licensing, none of which is mentioned in the announcement.

Buyers should treat speed claims as unverified until tested on their own cases.

Sources

• Seraphim Space: Zenithon raises 10M for AI world models — 30 September 2026
• Tech.eu: Zenithon raises 10M to build world models for extreme physics — 30 September 2026