Chery added that there are dynamics present that were not there six months ago. "The electric car is a megatrend and it equalizes the barrier to entry. This is accelerating and transformational and we have to deal with it."
And what about artificial intelligence (AI) which seems to be the hot topic right now?
Chery acknowledged that AI has proved transformational in the datacentre and is now moving to consumer equipment and edge devices where it will also have an impact; but he declined to use the word revolutionary. "AI-at-the-edge is an incremental improvement that will boost the digitalization of IoT," said Chery.
He said ST's strategy will be to address it through the company's STM32 microcontroller family. Initially by providing easy-to-use tools that will allow the creation and training of neural networks in software and subsequently to provide options to boost performance of those networks through optional DSP-like hardware accelerator cores on the microcontroller chips.
What ST does not plan to do is what a lot of startups have done; try and create the definitive neural processor as a full ASIC or application processor, Chery said.
He said that ST has AI experience from working on the fourth, fifth, and sixth generations of Mobileye ICs with AI embedded within them. While it represents a way of tackling otherwise hard-to-solve tasks it also raises a lot of questions about safety and where liability lies.
"I don't want to be engaged in a very application-specific processor again," he said, in an apparent reference back to the ST-Ericsson smartphone processor project that lasted from 2009 to 2013 and was the cause of years of financial losses at ST.
Next: In-house PCM but not MRAM