In partnership with

I hope you enjoyed the Conception X demo day roundup in our previous edition. We certainly enjoyed a good traffic bump from it, so that’s a good sign!

Today we’ve got a deep dive on one the startups unveiled at the event, and as usual, Premium subscribers get the full story.

But first:

• We’re still collecting your thoughts on the state of the pre-seed landscape in 2024 and we’ll be publishing the results in early January.

• That means you can still take two minutes to fill in our survey, whether you’re a founder or investor: Do it now! 🙌

– Martin

The future of presentations, powered by AI

Gamma is a modern alternative to slides, powered by AI. Create beautiful and engaging presentations in minutes. Try it free today.

Mater AI wants to heat up the hunt for electricity-generating materials

You might not think of heat as a waste product, but for people in the field of thermoelectrics, it’s often a wasted source of energy we could be putting to use to help meet our future needs.

A commonly-shared statistic is that around two-thirds of the energy we produce is converted into waste heat that disappears into the atmosphere. 

Thermoelectric materials have the ability to capture this heat and generate electricity from it. It’s already a valuable market thanks to materials like bismuth telluride

But as Mater AI CEO Nickel Blankevoort puts it, there’s still a lot of work to be done in this field:

“The beauty of these materials is that they can convert heat into electricity without any moving parts, so it's basically maintenance-free.

“However, the downside is that the efficiency is not yet high enough for these materials to be economically viable. So current applications are mainly in spacecraft, because there you need high reliability, no maintenance, and they can deal with slightly lower efficiencies.”

Expanding the market relies in part on finding more efficient materials, and that’s where Mater AI wants to step in. The London-based startup has built a “computational discovery engine” to help discover such materials. 

Rather than relying on chemists to experiment based on their own ideas in a trial-and-error approach, Mater AI is using AI and physics-based modelling to accelerate the process “from decades to minutes” as they put it in their Conception X demo day pitch last week

“There are almost an infinite amount of possible ways you can put molecules together, so looking at them one-by-one with human cognition is not the right way,” says Blankevoort.

“We set our desired material properties for an efficient thermoelectric material. And then we let our engine, which is trained on data sets, iterate over novel molecular structures until the desired properties are met.”

The long-term plan for the startup is to synthesise the materials they identify so they can sell them to customers. 

The story so far

The Mater AI story begins when Gatleen Bhambra, who has worked in a range of tech industry roles in recent years, met quantum physicist Chelsea Williams who is currently doing a PhD developing quantum algorithms to solve differential equations.

Looking for a startup idea to put this work to use on, Williams met Blankevoort on Y Combinator’s Co-Founder Matching platform. Blankevoort has a PhD in computational material science, and had previously gone through the Entrepreneur First programme but was still looking for the right way to move forward with his ideas for a startup.

Williams had developed an interest in venture science and was on the Conception X programme but needed a founding team to work with her on her own startup plans.

“Chelsea and Nickel realised there would be a really strong synergy between the fields of AI and quantum, and the field of material science,” says Bhambra.

“Then really quickly, I think that same week, we all jumped on a call and realised this is something we want to do, and we've been working together ever since.”

Now with a proof-of-concept product complete, they’re ready to move forward.

Next steps

Blankevoort says the next stage for the company is to ensure they can not only identify viable materials but also make sure they can actually be made in a cost-effective manner.

“We are looking into adopting algorithms that predict the synthesisability of novel molecular structures. And once that's incorporated into our engine, we are not only predicting materials with high efficiency, but also ensuring at the same time that these materials can be synthesised economically.”

Within the next two years, they aim to have synthesised a material that they can test and deploy in their first target market: data centres.

“They convert 98% of the electricity that goes in, into waste heat. So we have been engaging and speaking a lot to data centres and their hardware providers… Our route to market will most likely be through hardware providers that manage the entire heat landscape and cooling landscape of data centres.”

“We are very eager to make these materials a reality. They do exist, it’s a proven principle. Now it's just a matter of finding more efficient ones, and it's just making sure that we are picking the right direction in the vast universe of molecular possibilities.”

Subscribe to Premium to read the rest.