Intel and Tower, the wedding of the carp and the rabbit

Yole Développement (Yole) has recently published two technology & market reports, Imaging for Automotive and Computing and AI Technologies for Mobile and Consumer Applications and one Processor Quarterly Market Monitor. These highlighted some interesting views relevant to the acquisition of Tower semiconductor by Intel Corp.

Intel has subscribed to the “More Moore” (MM) technologies and markets since its inception; being founded by Gordon Moore itself is no small heritage and the reason why Intel’s embodiment of the law is intricately woven into its DNA. In recent years, Intel has made some interesting moves beyond its core business. In 2015 it famously acquired Mobileye, an automotive computer vision (CV) startup, for $15.3B and was heavily criticized for doing so. However, this venture became extremely successful and placed Intel at the center of the current robotic vehicle craze as well, as the highly automated ADAS L2 and L2+ cars.

For a few years, Intel has been under intense pressure from investors to reinvent itself. Not only has the revenue growth of Logic chips been below that of memory and “More than Moore” (MtM) devices, but its core products and markets have lost ground. This was primarily because new markets have emerged, like smartphones where Intel was unable to get market share, or new product categories like GPUs and FPGAs, multimedia chips, for which Intel had no proper market access or product alignment. Does this mean its “More Moore” strategy is doomed, and Intel is switching to “More than Moore”? Not at all. But despite the need for computing power increasing exponentially, the devices supporting this growth are also changing rapidly, so committing to the right opportunity is tricky. Having a prominent position in a previous technology cycle does not give any advantage in the current nor the next.

The previous technology cycles, whether we are thinking of personal computers or smartphones or internet infrastructure, have seen the emergence of specialized semiconductor companies, Intel for CPUs, SK Hynix for memory, Qualcomm for APUs, Broadcom for Analog. We could, therefore, ask why on earth would Intel buy a specialty foundry company like Tower? Much has been written regarding the open foundry business Intel wants to master, the looming talent shortages, and the geopolitical tensions forcing companies to grab as much of the semiconductor pie possible before others take a cut. But in doing so, is everyone maybe missing the point Intel is making? “More Moore” and “More than Moore” are now interlaced. They are serving the same emerging applications, the same customers, and therefore could be served by the same company.

In building a personal computer or an internet server, there are no compromises to be made except for the cost-to-performance ratio of the computing chip. In smartphones, things are a little more complex; the performance of the whole end-system is a compromise between the APU, the embedded memory, and ultimately the RF module, the cameras, and so on. It is interesting to note that Samsung, which managed to master both memory and APU, finally supplanted Intel at the top of the semiconductor industry. The trade-off for smartphones was indeed mainly between two types of “More Moore” devices, a compromise between computing and memory.

Nowadays, things are evolving quickly, and “More than Moore” devices are finally hitting prime time. This is already obvious in current smartphones, with multiple cameras and 5G connectivity, but also in the emerging applications, such as drones, mixed reality, autonomous vehicles, and industrial robots. The new compromise is extending to “More than Moore” devices. Our belief at Yole is there could be a mathematical linkage between the perception (sensor) or communication system generating a data stream and the computing power needed to handle this data. From what we have observed thus far in the realm of autonomous cars is that “computing power is evolving with the square of data”. Not only does computing power become a key limiting factor, but there is pressure all the way down to the performance of the perception, communication, and actuating devices. Even with unlimited computing power, robotic cars are not yet fully autonomous, nor are the cashier-less stores fully functional. There is now a “More than Moore law” that provides system-level improvements at a certain speed that the semiconductor industry is striving to fulfill.

The next “More Moore” solution – including both computing and memory – will therefore imply a compromise with many “More than Moore” devices, whether they are a power converter, an image sensor, or a photonic transceiver. The Semiconductor giants of tomorrow will have to speak both the language of the carp (More than Moore) and the rabbit (More Moore). The marriage of Intel and Tower indeed promises a lot of disruption, and we wish the best and many chip children to the happy couple.

About the author

Pierre Cambou, Yole Développement

Pierre Cambou MSc, MBA, is a Principal analyst, Imaging and Display in the Photonics and Sensing Division at Yole Développement (Yole).
Pierre’s mission is dedicated to imaging related activities by providing market & technology analyses along with strategy consulting services to semiconductor companies.
At Yole, Pierre is responsible for the CIS Quarterly Market Monitor while he has authored more than 15 Yole Market & Technology reports. Known as an expert in the imaging industry, he is regularly interviewed and quoted by leading international media.
Previously, Pierre held several positions at Thomson TCS, which became Atmel Grenoble (France) in 2001 and e2v Semiconductors (France) in 2006. In 2012, he founded Vence Innovation, later renamed Irlynx (France), to bring to market an infrared sensor technology for smart environments.
Pierre has an Engineering degree from Université de Technologie de Compiègne (France) and a Master of Science from Virginia Tech. (VA, USA). Pierre also graduated with an MBA from Grenoble Ecole de Management (France).

Related products

Towards 20 cameras in upcoming cars, imaging for automotive will reach $25B in 2027.

Computing and AI technologies for mobile and consumer applications 2021 - Cover

Penetrating everyday products will see the market for AI technologies for the consumer market reach $5.6B in 2026.

Application Processor – Quarterly Market Monitor BD

For the first time, the processor monitor is including FPGA, CPU, GPU, and APU including all the IDMs, Fabless companies, and Foundries in the business.

Status of CIS Industry 2021

The CMOS Image Sensor industry maintained significant 7.3% YoY growth in 2020, reaching $20.7B.