Yole CEO Jean-Christophe Eloy outlined five technology trends worthy of investigation and investment in the months and years to come...
Behind every strategic decision lies a wealth of market research and analysis. Since its creation in 1998, market research firm Yole Développement (Lyon, France) has been committed to looking beyond the obvious to provide a reliable barometer of global electronics industry trends. Acquisitions, partnerships, product launches, economic downturns and upturns, even specific customer requests — all raise questions and open new fields of investigation.
In an interview with EE Times Europe, Yole CEO Jean-Christophe Eloy outlined five technology trends worthy of investigation and investment in the months and years to come.
Quantum computing could open possibilities we never even considered in chemical research, financial services, health care, life sciences, manufacturing, and defense. Quantum computing is one of the key trends likely to have a long-term impact on the industry, and Yole’s customers are eagerly asking for guidance. What is the expected silicon content in a quantum computer? How will the new paradigm affect business in the next five to 10 years? Will it change the structure of the industry?
“In quantum computers, there is a lot of coding, cryogenics is huge, and silicon is nothing,” said Eloy. “The value is not in the silicon but in the system generating the cryogenics.” Many questions also surround neuromorphic imaging and computing, “which change the way we get and compute information,” he said. Neuromorphic approaches are not only game-changers technologically but could also change the value structure of the market. Yole is “trying to understand how the value flow is changing the supply chain and who will be able to do something in that area,” said Eloy.
Knowles Electronics, for instance, identified a turning point in edge audio processing with the emergence of artificial intelligence and took actions to make its products more powerful and suitable for AI. With the Google Pixel 4 smartphone, the Illinois-based company saw an opportunity not only to place microphones but also to add a processing chip, increasing its value in Google phones by a factor of four or five.
“If you combine the microphone with the ability to process the data and do specific things around the data with added value for the mobile-phone or car maker, you move from the few-cent sensor business to the few-dollar business,” said Eloy. Companies like Knowles then multiply the value of their selling and climb the value chain.
Auto manufacturers are not born equal. Tesla, founded in 2003 by eclectic Elon Musk, calls for “disruptive ways of working.” Because it has no business history, it can map out technical and supply chain strategies on a clean slate. As a result, Tesla “is able to jump on solutions that are years ahead compared with other companies,” said Eloy. Tesla is agile; it moves and adapts quickly. In contrast, traditional automakers such as Toyota, Renault-Nissan, and BMW have years of investment in technologies and long-term relationships with suppliers, making it “almost impossible for them to restart from a white sheet.”
Driven by the adoption of cloud-based technologies, data centers are proliferating in size and number around the world. The total number of hyperscale data centers — housing data, enabling entertainment, connecting consumers everywhere, and powering and cooling infrastructures — now exceeds 500.
Given those trends, Yole has received repeated customer requests for more intelligence on data centers. But for now, “everybody is driving in the dark,” said Eloy. “Data centers are giving us a headache.” Over the years, Yole’s analysts have closely observed the data-center market, including ups and downs in investments, movements in the supply chain, and technology choices. But without access to the parts, analysts can’t do the reverse engineering to understand what is really used. Such opacity is due to the direct relationships between the companies designing and ordering the data centers (e.g., Amazon, Google) and the companies manufacturing them. The absence of intermediaries in the chain makes it difficult for Yole to gather information and build its own analysis.
With context-aware technologies, personal devices such as smartphones and home assistants can now collect and process raw contextual data about their users’ surroundings, preferences, and behaviors in situ. Mobile context awareness enhances the quality of service by providing more relevant responses, but it also raises security and privacy concerns. A wealth of personal data and private information is accessible, and “nobody is really sure of the way these devices are managing the data,” said Eloy.
“I am surprised to see, every quarter, problems of data leaks in all the big applications,” he added. Sensitive data immediately becomes vulnerable to cybersecurity risks, but customers have had limited reactions. “This is not acceptable.”
In 2018, Europe came up with the General Data Protection Regulation (GDPR), which is still considered the most comprehensive privacy law. On the other side of the Atlantic, Apple preaches privacy and has made it a powerful marketing asset. Today, Eloy said, companies simply say, “‘Sorry, we will do better next time,’ [but] my feeling is that there will be an amplification of this movement in the next few years to make sure your credit card is not hacked just because you are ordering something from your mobile phone or your home assistant.”
Data leakage over the cloud is also a major concern for corporations and governments. In response, said Eloy, “the movement to the edge is huge. It will never stop, which means there will be more and more closed systems to protect data as much as possible.”
The transition from the cloud to the edge is indeed accelerating. Despite always-on power consumption and data-processing constraints, edge computing is emerging, placing the calculation at the system level.
As Moore’s Law fast approaches its physical limits, the industry is under pressure to develop a new type of architecture. The trend toward systems-on-chip is decreasing because SoCs “cost a lot of money,” said Eloy.
Chip partitioning and chip stacking are ways to extend Moore’s Law and yield power, performance, area, and cost benefits, he added. It is all about having more processing power without “moving the whole chip to the 5-nm node.”