This story is syndicated from the premium edition of Preseed nowa newsletter that takes a closer look at the product, market and story of the founders of UK-based startups, helping you understand how they fit into what’s happening in the wider world and the startup ecosystem.
The newfound excitement surrounding AI’s potential as we storm into 2023 brings with it concerns about how best to process all the data needed to make it work. This is far from a new challenge, however, and next-generation AI chips are being developed in labs around the world to address the challenge in a variety of ways.
One of the first startups we ever covered at PreSeed Now takes a ‘neuromorphic’ approach, influenced by the human brain. Coming from a different direction, a brand new Newcastle University spin-out is mentioned Mignon (so new in fact that there is no website yet).
Buy your tickets for TNW Valencia in March!
The heart of technology comes to the heart of the Mediterranean
Mignon has developed an artificial intelligence chipset that, according to CEO Xavier Parkhouse-Parkerhas “on the order of 10,000x performance improvements over alternative neural network-based chips for classification tasks”
Classification is essentially the process of figuring out what the AI is looking at, hearing, reading, etc – the first step in understanding the world around it, regardless of its use. Mignon’s chipset is designed to be used in edge computing as a “classification coprocessor” on devices, rather than in the cloud.
In addition, Parkhouse-Parker says Mignon’s chipset can also train AI models on the edge, meaning the models can be optimized for the specific, individual environments in which they are used.
A proposition proposal
What Mignon says gives his technology an advantage over the competition is based on a less resource-intensive approach proposition logic.
“Neural networks, the dominant algorithm in AI and machine learning today, typically require many layers of increasingly resource-intensive computation to be performed. They can take a very long time and take a huge amount of energy to train and deploy, and they also exist as a black box; you can’t explain why the algorithms came to a certain conclusion,” says Parkhouse-Parker.
“Mignon is based on an algorithm that can be executed in a single layer, using propositional logic, while maintaining accuracy, but allowing calculations to be performed much faster, using much less energy.”
And when it comes to market launch, Mignon could also have a big advantage.
“The investments and commercial scale required for success in the semiconductor industry are significant. One of the biggest challenges for many other competitors in this industry is that they rely on non-standard or ‘exotic’ features that are not easily scalable within today’s semiconductor manufacturing ecosystem,” says Parkhouse-Parker.
Instead, Mignon’s chipset uses a kickstand CMOS manufacturing approach, which means mass production is much easier.
How can it be used?
Edge AI has already made a remarkable difference in consumers’ lives. Just look at how companies like Apple and Google have put AI chips into their smartphones to perform tasks like face and object recognition in photos or audio transcription locally, increasing privacy and speed while reducing data transfer costs.
Parkhouse-Parker says Mignon could eventually make a difference here, along with in next-generation ‘6G’ telecom networks, where signal processing could be optimized by AI
But the first market they’re looking at is industrial spaces where connectivity and energy resources are low, but there’s a need for powerful AI classification.
And while the technology isn’t ready yet, Parkhouse-Parker says Mignon is working on another selling point that will enable his offering: “explainable AI.” That is, transparency about how and why AI made a certain decision.
To give a current example, if you ask OpenAI ChatGPT to explain a concept to you, you don’t understand why it comes up with the specific answer it gives. You just get an answer based on the path it took through its sea of data in response to your prompt.
In an industrial environment, where AI might be making business-critical decisions, or decisions with security implications, it would be very helpful to be able to look back and see how the AI came to the conclusion that this was the case.
“With neural networks, all inferences are made within a black box, and you can’t see how or why this node connects to this node, or how things are calculated. With Mignon, because it is based on propositional logic, a researcher can look inward and see exactly where a decision was made, and why, and what led up to that point,” explains Parkhouse-Parker.
Mignon wants to enable this form of accountability via software, which could be attractive in areas such as medicine, defense and the automotive industry.
Their research into taking the Tsetlin machine and putting it into computational hardware caught the attention of the deep tech enterprise builder Cambridge future technologywhich – among other things – also works with GitLife Biotech and imitatepreviously featured in this newsletter.
Parkhouse-Parker (COO of Cambridge Future Tech) has been developing a commercial proposition for Yakovlev and Shafik’s research since the spring of last year. He has taken on the role of CEO at Mignon as it leaves university.
Enter the market
First on the to-do list for the new startup is further refining its technology with the development of a ‘generation 2’ chipset before they bring it to market.
“While we have fantastic performance gains, and it’s actually quite remarkable, all of this was done on the 65-nanometer node, which is an old technology and should mean worse performance gains because the transistors are actually bigger, which is what we truly remarkable,” says Parkhouse-Parker.
“We think when we move to a 28-nanometer node, whatever numbers we have, the benchmarks at this scale will be significantly larger.”
Commercial validation is of course another important step. The ultimate goal is to partner with fabless chip companies to build the Mignon technology into a commercially available system-on-a-chip. Mignon has a number of hires planned for the near future to get there.
Investment plans and future potential
Parkhouse-Parker expects the spin-out process to be completed by March this year, after which they will formally open a £2.55 million round of funding.
This will be used to expand the team, develop, test and manufacture the next generation chipset, and gain commercial validation in a number of industries. Software to enable AI development on the chipset is also an important part of the roadmap.
Ultimately, Parkhouse-Parker wants Mignon’s combination of power-efficient performance and widespread compatibility to usher in entirely new opportunities for AI
“What Mignon is doing is opening up an opportunity for a truly completely new world of devices that people haven’t even thought of yet. Think about the opportunities there would be with product people like a Steve Jobs or a Jony Ive who could use this and get wild with the potential. I think there really is a whole new world of possibilities.”
The big “bump”
There is no clear path from where Mignon is now to that future. Aside from the additional development work to fine-tune the chipset, it will require a mindset change from the people building AI applications.
“The big ‘hump’, as one of our consultants puts it, is that it’s a new way to engage with artificial intelligence,” says Parkhouse-Parker. “The transition between neural networks and Tsetlin is not incredibly significant, but it requires a slight difference in mindset. It may require new ways of thinking about how to design artificial intelligence problems and how to market these things.
“There’s a great community being built around this already, but that’s one of the biggest challenges is building a Tsetlin ecosystem and turning things that are neural networks into Tsetlin.”
But despite these challenges, Parkhouse-Parker believes Mignon’s vision is very much achievable.
“Several orders of magnitude improvement warrants a look at something new, novel and exciting.”
The article you just read is from the premium edition of Preseed now. This is a newsletter that delves into the product, market and story of startups founded in the UK. The goal is to help you understand how these companies operate fit into what is happening in the wider world and the startup ecosystem.