FiveAI very recently announced themselves as the latest start-up on the self-driving car scene, with an investment of $2.7m. Its CEO is also chairing a fast-growing trade association, NMI, which is supporting deep tech investment in the UK and hosting a major event on the subject, Future World, on September 15-16 in London.
We spoke to FiveAI CEO Stan Boland to find out why there’s a renaissance in deep tech companies in the UK and hear his own company’s unique proposition.
Stan Boland: Chairman of NMI, CEO and founder of FiveAI
FiveAI represents Stan’s first venture into the AI marketplace, though he previously founded, ran and sold two highly successful European telecommunications start-ups – Element 14 (a fixed line communications silicon and software company acquired by Broadcom for $640M in 2000) soon followed by Icera, a cellular silicon and software company, that grew to over 300 employees by the time it was acquired by NVIDIA in 2011 for $367m.
What do you mean by deep technology and who is NMI?
“It means fundamental engineering that relies on mathematics, physical sciences and computer science to deliver core capabilities that allow applications to marketplaces to be built. So that includes technologies like machine learning, AI, cryptography, semiconductors, AR/VR and the like. In fact NMI (National Microelectronics Institute) started life as a trade association for the UK’s semiconductor and electronics industries – we have about 100% of that community as members – but more recently has forged new trade networks in adjacent deep tech segments, one in Internet of Things Security (the IOTSF) and one in automotive engineering (AESIN).”
And the event in September is to promote investment in the technology space?
“Exactly, it’s called Future World and will be held on September 15-16 in London’s Truman Brewery. We’re bringing together start-up entrepreneurs, established firms, venture capitalists and analysts to create a buzz around the new wave of deep tech investing. We’re aiming for it to be to deep technology what Slush is to marketplaces.”
Stan struggles to see an industry that won’t be affected or transformed by AI. “It will affect every sector and probably every job”, he said.
And despite admiring what Google and Facebook are doing – as well as the other companies investing heavily in AI – Stan feels the “true innovation comes from start-ups” – he hopes his own!
So after your proven successes in Element 14 and Icera, why AI now? Stan describes AI technologies as the next “rotation”:
“15 or 20 years ago, telecoms infrastructure was at its peak level of research and investment. Then, as that was built and wireless technologies matured, consumer devices became the place to focus, the attention was on that paradigm shift from 2000 to 2012 or so. We’ve gone through 3G and 4G with 5G on the horizon, and even the IoT might be reaching its peak of R&D. But, after the 20 years of darkness, over the last 5 years there have been staggering breakthroughs in AI. That’s helped by the fact we have the compute resources available, libraries and tools exist and work is widely published. AI and ML is the new paradigm in which investments are being made and where the best people are carrying out research”.
But FiveAI is also Stan’s first venture into the auto-industry. And in a space where Google’s car dominates the headlines – and where many more manufacturers besides are doing their own research – it might appear an intimidating market to crack. Stan’s company, however, have plans to take a fresh approach to how their software is architected and then prove its effectiveness, drawing on some of the latest AI research. Rather than build their own car, like Google or Tesla, they will license their software to integrate into the OEM’s cars.
And Stan tells us there’s another crucial difference between FiveAI’s proposition and that of other autonomous car developers. “Contrary to the need for centimetre-accurate prior 3D mapping that is required for Google’s car, FiveAI’s software, together with a set of sensors and cameras, will deliver much better accuracy as it goes; that should enable its vehicles to navigate without such detailed prior maps. Since there are 37 million kilometres of road across the planet and since Google’s car has clocked up well over two million kilometres of driving, you might think Google is well on the way to delivering driverless cars, but in fact it has only done so over 3,000 kilometres of mapped roads. So there is a huge distance to go. Once full up and running, FiveAI-fitted cars should be able to drive into new, complex road environments with just a simple navigation map and with just a limited knowledge of the environment beyond it”.
That stage is a while off though, Stan assures me. “We are currently in phase one”, he explains. “We need to make sure that every layer of our software and all our components work and then bring that up on a test car. Building in system resilience, writing software to international standards, generalizing complex situations that represent the long tail of possible scenarios and working with OEMs to fully integrate the software and sensors into the vehicle would be phase two”. During these early phases, Stan expects the company to expand significantly: “right now we have six employees – by the end of next year I would expect us to have around 20 and by the time of possible service launches in 2019, we’d expect to be nearer 80 employees”.