China tells Europe it is better than it thinks, as industry debates AI arms race
Chinese giant Alibaba told a packed-out room at Electronica 2018 in Munich, in Germany, yesterday that Europe is well placed in the emerging tech arms race against China and the US, specifically for development of artificial intelligence (AI) technologies.
UK universities and German corporations, notably Bosch and Siemens, have the research side cornered, it suggested, during a CEO rountable session at the Munich event, even if Europe struggles to fund and foster startup innovation to more widely commercialise AI applications.
Xianfeng Ding, president and chief executive at Alibaba, said Europe has the best academic and corporate research institutes, specialising in AI development, of any region. A rather disbelieving panel, comprising chief executives from Europe-based semiconductor companies Infineon Technologies, NXP Semiconductors, and STMicroelectronics, among others, had suggested already in the session Europe is lacking in certain aspects of AI development.
Ding knows Europe’s capabilities well, he explained. He worked previously at STMicro and Bosch in Europe; his brother worked for Infineon and NXP, he added, gesturing to his fellow panellists. The room laughed. He went on to praise, in high fashion, the research platform Europe has established for digital technologies.
“In the academic area, Europe is very strong,” he said, referencing the science emerging from the two ‘Oxbridge’ universities in the UK, as well as from the Technical University of Munich. “Europe is number one for AI research. It is ahead of the US and China, in my opinion.”
He also singled out the four top private-sector research centres for AI, as Microsoft, Siemens, Bosch, and NEC. “They are the top four corporate research institutes, and two are from European companies. So in terms of corporate research, Europe is also very strong. I believe European companies are top of the world in terms of IoT,” said Ding.
“And actually, the mission of IoT is to digitise the world. The mission for AI is to data-mine, after the world has been digitised. And Europe is the number one continent for that. Siemens is the number one company. These companies, like Siemens and Bosch, started on this road 10 years ago, way ahead of China and the US.”
“My view is Europe is way ahead in terms of academic and corporate research. What is missing in Europe is venture capital, and financing; the startup movement is not as strong as in the US and China.”
Beside him, Reinhard Ploss, president and chief executive at Infineon Technologies, suggested Europe has a vested interest in industrial AI development because of the strength of, and the importance to, its automotive sector, in Germany at least. “The reason AI is strong in Europe is we have strong backbone is automotive, which has a strong interest in AI – because AI will have a big relevance in how that sector moves forward,” he said.
European governments are pushing AI hard, as a vehicle to future industrial growth. However, Europe finds it hard to move away from the science of it, he agreed. “Germany is not strong in high level of digitalisation; it is strong in embedded, but AI might open a new window to get into that higher area,” said Ploss.
In a wide ranging debate, which also discussed challenges in Europe with funding grassroots innovation, and with the regulatory issues around underlying infrastructure as a springboard for innovation, Ploss said there is a hard core of academic research into AI in Europe, but a challenge to translate that into commercial expertise.
“The question is how to industrialise this knowledge – for government and industry to link the science and make it happen. It is not a matter of AI alone, but how to fuse it [in commercial applications], and Germany has some way to go,” he said.
Generally, French-Swiss firm STM was more bullish. AI is alive and kicking, already, and Europe is well placed, it said. “AI at the edge is an incremental innovation. It is not the radical innovation everyone talks about,” commented Jean-Marc Chery, president and chief executive at STMicro, making the point its application in industrial manufacturing processes, to ascertain and predict faults, is compelling already.
Does it consider Europe to be at a disadvantage, compared with the US and China? “I don’t think so. You have STMicro and NXP, here, and some of the best European companies. We are leading in IoT. WE offer all the features and processing, and power and activation. These [companies] are excellent leaders in the world.”
But Kurt Sievers, president at Dutch firm NXP, and a fixture on the first day at Electronica 2018, with a keynote and press conference besides the CEO session, sees the AI race rather differently. “Compared with China and the US… I am not optimistic. The US has been leading from a scientific perspective, and China from an implementation perspective. In Europe, we have to fight our role,” he said.
Ploss rejoined, suggesting European innovation is concentrated in well established multi-nationals, and lower-level innovation is stymied, compared with the US and China. “In Germany, lots of innovation is based in large companies. We need to accelerate understanding of how to make the best use case out of AI, and bring it together with analytical programming to have real decision making… Nothing matters if you don’t have use cases to go to,” he said.
Alexander Kocher, president and managing director at Elektrobit, a 3,000-strong automotive software firm based in Germany, acquired by Continental in 2015, suggested AI is well used in smaller European outfits, too.
“We’ve been doing it for years,” he said, referencing image and speech recognition, and navigation technologies in cars. “Even for a small sized company, with limited financial investment.” The Continental deal has seen it move the technology on, he explained, in line with the demands from the automotive market for functional safety applications.
Walden Rhines, also on the panel, heads up Mentor Graphics, the US software design acquired by Siemens in 2017 for $4.5 billion, as president and chief executive. He explained the German firm’s strategy, to go west for its AI talent.
“The part of Siemens that acquired Mentor Graphics is the software business, based in the US, which is a little less than five per cent of the company. Mentor Graphics was a $1.4 billion revenue company, dedicated to design implementation, principally. The effect of AI is massive for Siemens, because of its role in Industry 4.0, and in dig twins – which are based on massive amounts of data, and environments for simulation,” he explained.
Momentum in the market for AI technologies is building, he noted. “When you narrow it down to the small piece I run, which is all in the area of automation and design, a few years ago there were AI development projects here and there, and now you can’t keep count.”
It was a good fit, he reflected; Mentor Graphics has seen business jump on the acquisition, in a way companies rarely achieve so quickly with new owners.
“We saw an advantage in the merger, which has showed up in our results. Normally, you see companies go through a period of slowdown [after a merger]. We didn’t. We have broken records for all-time growth. And we can attribute that to the financial resources and credibility of Siemens. Because we weren’t acquired by unfriendly resources. Siemens went out of its way to have a process to be sure we could go together, and how we could be separate.”
Sometimes established power pays off. Ding gave the example of China based startup Horizon Robotics, a darling of the AI scene in his homeland, lead by a TUM graduate, and a close familiar of Bosch.
“Bosch visited a lot of times. And actually Bosch did a lot of work in that [field] for 20 years already. The guys there [at Horizon Robotics] are smart; the guy leading it was from Munich. It had a lot of money, but it hasn’t accumulated 20 year [of expertise] like Bosch, so it is aiming at a niche market,” he explained.
He restated his view. “Europe has all the assets to be a leading continent in research and industrial.”
For Sievers at NXP, Europe’s inherent conservatism, and well judged pragmatism, brings certain opportunities for the region, however. It will provide the safeguards, to make the runaway capitalist innovation of the US and China commercially viable, ultimately, for end users. It has taken a lead on data protection, already, he noted.
“I believe we can make difference by providing security and privacy protection to AI applications, and functional safety in driving applications, because that protection will be the tipping point between consumer acceptance and a loss of trust. This is where Europe will play a leading role.”
Ploss said the same; the cautious European approach is appropriate for tackling issues of safety, security and privacy. “Of course people are afraid. But that’s not a bad thing – that they voice concerns. In Germany, we are often in awe of the problem, but [that sets us] to find the solution,” he said, making the point deliberate and careful problem solving is the only way.
Sievers noted Europe’s lead on data protection, by passing ‘general data protection regulation’ (GDPR), which is being re-modelled in other regions.
“In Germany, worked heavily on GDPR, which was criticised initially. But the US administration is looking at it pretty closely now – it won’t be the same, but it will be adopted as a concept. We seem to export security from a conceptual perspective…. And with GDPR, we have made a good step forward,” he said.
Ding said Alibaba is the first cloud company in China to comply with GDPR. “We think that’s very important – to encrypt the data all the way from the sensor to the cloud,” he said.
Sievers talked in a separate keynote about the importance of edge computing to bring data processing closer to the action, and to remove the security risk of forwarding it to the cloud for processing.
“The difference is where lots of this deep data mining is moving to the edge – on the industrial work floor, and in hospitals and cars, and in the home, too. The breakthrough will come when we get good machine learning algorithms at the edge…. That goes big time hand-in-hand with security, because it means less exposure. Those aspects are super relevant from a European perspective,” he said during the panel session.