By Junko Yoshida

What’s at stake:
TinyML in embedded systems can be implemented many ways, often by leveraging beefed-up MCUs, DSPs, AI accelerators and Neural Processing Units (NPUs). The lingering dilemma is how best to develop embedded systems with machine learning that could fit in the budget of TinyML.

Almost every new technology overheats its industry’s imagination, followed by an announcement flood, promising new tools, software and hardware – all of which fuels dreams of rapid market growth and big volume sales.

Then, reality.

TinyML has reached this moment.

By Junko Yoshida

What’s at stake:
The concept of software-defined vehicles for whole-vehicle architecture is the automotive industry’s hot topic. But SDVs, for most OEMs, are still in early development. Nonetheless, “software-defined sensors” appear to have traction among carmakers. Why?

Software-defined sensors are going commercial way ahead of software-defined vehicles. This is because OEMs are now required to develop cars compliant with federally mandated Automatic Emergency Braking (AEB) rules, including Pedestrian AEB in low light.

Carmakers must pass tests of minimum performance criteria and meet a National Highway Traffic Safety Administration (NHTSA) timeline.

OEMs are seeking an answer in “software-defined” sensors, largely because it might enable them to pass AEB tests without adding new sensor hardware such as thermal cameras or lidars.

By Junko Yoshida

What’s at stake:
Ai is disrupting the electronics industry. Increasingly diversified AI workloads are triggering seismic changes in the architecture of chips, boxes and data centers. Synopsys explains how AI is shortening PCIe spec cycles and discusses the role of next-generation interconnects in AI-driven data centers. 

The tech industry understands AI’s voracious appetite for more data, computing power and memory, and is coping — sort of. But so far not discussed enough are “interconnects” inside a box that have to migrate outside the box.

“In the world of interconnects, we are beginning to hit the laws of semiconductor physics,” said Manmeet Walia, executive director, mixed-signal PHY IP, Synopsys, in a recent interview with the Ojo-Yoshida Report. In contrast, “with compute, you can still go faster by leaning on Moore’s Law. Or, if you can’t go any faster, you can start parallelizing processing.”

Interconnect speed is now clearly trailing compute, according to Walia. Worse, doubling the bandwidth of an interconnect – from 64 gigabits to 128 gigabits per second, for example – does not just double complexity. It introduces “an exponential increase in complexity,” he noted.

By Junko Yoshida

What’s at stake:
The contrast between ST and Wolfspeed couldn’t be starker. Did Wolfspeed, focused intently on leading the transition from 6- to 8-inch SiC wafers, underestimate Chinese wafer makers? Or, too busy satisfying multi-year wafer contracts with SiC device vendors such as Infineon and Renesas, did Wolfspeed fail to see that profitability in the business has already moved on from wafers to devices? We pick the Yole Group’s brain to learn what ST has done comparatively better.

STMicroelectronics last week unveiled its plan for a new 8-inch SiC manufacturing facility in Catania, Italy. That site will become an integrated hub for ST’s comprehensive SiC operations, from wafers to testing and packaging devices.

This move is monumental.

Above all, it paves ST’s path to become, over the long run, a genuine leader on the SiC market.

By Junko Yoshida

What’s at stake:
Why are incumbent automakers flocking to China? Gaining a share in China – now the world’s largest automotive market – is an obvious reason. But there’s another, bigger reason. They want Chinese partners.

Auto OEMs in the West know they need the nimbleness and daring necessary to build cars at a 12-18-month design cycle, faster than the current cycle of several years.

In short, they want to copy China.