中文 English

Big Growth Areas: Connectivity, AI, Reliability

Where the chip industry is heading this year, and where the biggest fallout will occur.

popularity

Connectivity and artificial intelligence (AI) will be the biggest drivers for 2020, with an emphasis on improved reliability across all areas. New standards, new applications, and new pressures being placed on old technology will created boundless opportunities for those ready to fill the need. Of course, there will also be a lot of carnage along the way, and we can expect to see a lot of that in the AI space this year.

Connectivity
New standards and devices will be hitting the airwaves in 2020. WiFi 6 (also known as IEEE 802.11ax) and USB 4 were launched into the market in 2019 and the Bluetooth SIG is expected to launch Bluetooth Low Energy (BLE) v5.2 this month. We also can expect to see a lot of progress in 5G.

“Semiconductor and equipment vendors have been introducing a range of chips and equipment for WiFi 6,” says Jo Jones, technology communications manager for Imagination Technologies. “It will trigger a major refresh of mobile phone connectivity in the first half of 2020. The benefits of the new standard include better throughput and robustness in dense environments and the potential to dramatically reduce battery drain. As equipment manufacturers become more familiar with these technologies, Wi-Fi 6 will quickly become mainstream. With features specifically designed for low power markets such as wearables, Wi-Fi 6 will open up new markets and opportunities for Wi-Fi.”

Then there is BLE. “This is not just another incremental update but a revolutionary step,” continues Jones. “One of the key features in the BLE v5.2 is Broadcast Audio. This will enable audio source devices (like TVs) to broadcast one or more audio streams to an unlimited number of devices, such as headsets and hearing aids. This means you’ll no longer have to stare at a silent TV screen at the gym, in the doctor’s waiting room, or in other public venues. TVs will be capable of broadcasting audio over Bluetooth, and multiple users will be able to access streams at the same time via their Bluetooth headset, even listening to the broadcast in their preferred language.”

Peripherals also will get a boost from USB 4. “The way we connect everyday devices, such as laptops, desktops, smartphones, to peripheral devices including monitors, virtual reality headsets, docking stations, external hard disk and solid-state devices, is evolving,” says Amin Shokrollahi, CEO for Kandou. “The adoption of USB4 with support for data rates up to 40Gbp/s over a cable and backward-compatibility with USB Type-C protocols such as USB3.2, DisplayPort and Thunderbolt 3, will drive a new generation of USB-enabled products.”

A Keysight think tank believes that “advanced applications related to 5G will explode in 2020, using higher frequencies and smaller geometries. In 2020 the industry will move from a small group of early movers, who have commercialized initial 5G networks to a global community in which multiple operators on every continent and in many countries will have commercial 5G networks. This will put pressure on networks, revealing new data center and network chokepoints.”

One of those application areas is automotive. “On the technology front for vehicle to everything (V2X), the battle is still ongoing between dedicated short-range communications (DSRC) and cellular V2X,” says Tom Wong, director of marketing for design IP at Cadence. “DSRC had a head start in Europe with VW backing it and putting it into practically every VW made in 2020. However, the groups supporting cellular V2X will not give up. They will continue to push and lobby to retain a seat at the table, but only DSRC will launch in 2020. Cellular V2X will have to wait until 2021 or 2022, as there is really no 5G infrastructure in deployment to support it in 2020. Even with the rollout of 5G in China, it is unlikely that cellular V2X will be deployed there since there are still unsolved challenges in making V2X technology viable.”

Another communications standard that is not talked about quite as much is LPWAN, but this could have a big impact on IoT devices. “Asset tracking will be the killer app for LPWAN,” says Hardy Schmidbauer, director of wireless products and IoT at Semtech. “New systems enabled by new indoor asset tracking solutions and improved ROI of indoor/outdoor solutions with much lower device and infrastructure cost will start to drive significant volume in 2020. Asset tracking applications will move ahead in total volumes over 2019 highest volume use cases of smart metering and smart building.”

Some of this is being driven by an increased interest in space travel. “Satellites will play a significant role in LPWAN going forward,” adds Schmidbauer. “There is flurry of activity in the satellite space around LPWAN. Renewed activity is driven by small, lightweight satellite designs and new low-cost deployment services. In addition to traditional players, there are around 20 new startups in the low-cost Low Earth Orbit space with satellite designs around the size of a basketball. Companies in the sector are projecting that world-wide coverage can be achieved for a cost of approximately $20 million to $25million.”

There is a larger trend going on, as well. “We are starting to see a need to move from the device era to an era of connectivity,” says Joseph Sawicki, executive vice president for IC EDA at Mentor, a Siemens Business. “This is evident in areas like IoT, industrial IoT, smart city, and autonomous driving. This means that verification needs to move to verification and validation of the digital twin — from the multi-chip interfaces into mechanical and real-world data — to ensure these complex systems work together.

Keysight agrees. “Digital twins, or the concept of complete replicate simulation, are the nirvana of design engineers. In 2020, we will see digital twins mature and move to the mainstream as a result of their ability to accelerate innovations. To fully realize the technology’s benefits, companies will look for advanced design and test solutions that can seamlessly validate and optimize their virtual models and real-world siblings to ensure that their behaviors are identical.”

There is also a significant change happening in the business models associated with connectivity. “LPWAN and IoT devices deployed under a traditional connectivity model will drop below 10% in 2020,” says SemTech’s Schmidbauer. “A connectivity business model or paying X dollars per month per device connected to a network is the most familiar business model in the communications industry, though the majority of enterprise and volume use-cases for IoT prefer and are selecting business models rather than connectivity only. The driving factor behind non-connectivity business models are service-level agreement (SLA) requirements, access and ownership of data, the demand for complete solutions, and edge requirements.”

Artificial intelligence
Artificial intelligence (AI) and machine learning (ML) continue to be the hottest trend in the industry, creating a boom for the semiconductor and EDA industries. “There is a slew of new semiconductor startups, as well as established companies, developing AI/ML-dedicated ICs and IP,” says Mentor’s Sawicki. “We’ve been tracking fabless company venture funding for many years. In 2019, venture capital invested $860 million into AI/ML semiconductor startups, which means over the last 7 years, the total venture investment in the AI/ML semiconductor sector now exceeds $2.5 billion — well over 4X the investment in the next largest market segment, high-speed communications and 5G wireless.”

Another way to gauge interest in a subject is by looking at papers in conferences. “The research and application of AI/ML architectures and solutions will continue to accelerate,” says Harry Foster, vice chair for the 57th Design Automation Conference, and chief scientist for verification at Mentor, a Siemens Business. “In fact, a simple metric to support this claim can be measured by examining the number of AI/ML architecture and system design papers that were recently submitted for review to the 2020 Design Automation Conference (DAC). In 2018, 56 AI/ML architecture and system papers were submitted for review. That number grew to 92 submitted papers in 2019, and a jaw-dropping 194 submitted papers in 2020. In other words, AI/ML architecture and system design research is growing at an 86% CAGR. Perhaps this should not be too big of a surprise, considering that McKinsey & Co. recently reported that the AI semiconductor total available market will grow at 5X the rate compared to the non-AI semiconductor market.”


Fig. 1: Startup funding trends. Source: Mentor, a Siemens Business

The are plenty of application areas driving this development. “Three themes will shape the advancement of AI — automation, natural language processing (NLP), and trust,” says Sriram Raghavan, vice president for AI at IBM Research. “NLP will play a key role in enabling AI systems to converse, debate, and solve problems using everyday language. And with each of these advances, we’ll see more transparent and accountable practices emerge for managing AI data through tools ranging from explainability to bias detection.

Data becomes increasingly valuable. “AI’s need for data can pose a problem for some businesses and organizations that have less data than others,” continues Raghavan. “That doesn’t mean they can’t count on the support of AI. During the coming year, more AI systems will begin to rely on ‘neuro-symbolic’ technology that combines learning and logic. Neuro-symbolic is the ticket to breakthroughs in technologies for NLP, helping computers better understand human language and conversations by incorporating common sense reasoning and domain knowledge. Such breakthroughs will soon help businesses deploy more conversational automated customer care and technical support tools, while requiring much less data to train the AI.”

People at Arm believe that “IoT will help to commercialize AI in 2020, with IoT-enabled companies using physical real-time data from IoT devices to advance their AI business goals. And as the amount of data collected from IoT devices grows and machine learning is deployed in devices and the cloud, the machine learning models will need to be updated regularly to deliver better results. We foresee the development of technology to create ‘living models’ that are constantly updated and improving AI-driven business outcomes.”

But there are barriers. “To trust AI, these systems have to be reliable, fair, and accountable,” adds IBM’s Raghavan. “We have to ensure that the public can be certain that the technology is secure and that its conclusions or recommendations aren’t biased or manipulated. During 2020, components that regulate trustworthiness will be interwoven into the fabric of the AI lifecycle to help us build, test, run, monitor, and certify AI applications for trust, not just performance.”

We also can expect to see another side of this technology emerging. “Competitive, differentiated AI hardware is the result of a long-term, intensive R&D process,” says Imagination’s Jones. “This marathon not only creates complex, high-value products, it also creates a strong patent portfolio, which can act as a landmine to competitors. And like a marathon, innovation can get harder over time if you’re not used to it or equipped for it. Consequently, not all companies can run the marathon, with many start-ups having little to show for their VC funding when the money finally runs out. Therefore, 2020 will be the year that the number of AI hardware start-ups will start to reduce.”

Indeed, 2020 could be filled with a lot of disappointments. “Large numbers of companies will finally arrive with silicon and real results with real power and costs will be measurable for the first time,” says Geoff Tate, CEO of Flex Logix. “There are market segments, but within a given market segment it is easy to move between architectures, given the high-level description of neural network models. So we will see the onset of the culling of the herd as investors stop funding those companies that cannot measure up.”

AR/VR
Arm continues to believe that “VR headsets and AR goggles still have a niche, futuristic feel to them, and our recent research shows that fewer than 1 in 10 now want a VR headset as a gift. But 2020 will be the year that changes. VR will move into the mainstream, becoming commonplace in homes, businesses and educational institutions. New untethered all-in-one headsets are coming to market, and compelling, can’t-miss experiences are ahead this year, including VR sports from the Tokyo Olympics and highly anticipated gaming content. And don’t write off AR. Major tech brands are planning to launch AR smart glasses this year, and we’ll see people wearing smart glasses in public for the first time.”

Automotive
Some predictions don’t require a crystal ball. “One of these is the evolution of safety technologies in automotive,” says Jones. “The Euro NCAP (New Car Assessment Program) already has published a roadmap of what features will be required in order to grade new vehicles, and this tells us a lot of what we can expect to see. In 2020, specifically, there will be a focus on safety between vehicles through vehicle-to-everything (V2X) communication, allowing vehicles to exchange safety and real-time environment information. Another technology that will come to the fore is driver monitoring, which is rapidly maturing thanks to advances is facial analysis that detects how alert and attentive the driver is.”

The folks at Keysight agree. “The quantity and sophistication of sensors deployed in vehicles will increase in 2020, but fully autonomous vehicles will require more ubiquitous 5G connectivity and more artificial intelligence. The first C-V2X network will hit the streets in China, but they will be operating on an LTE-V network until 5G Release 16 evolves the standard.”

And safety is coming under increased oversight. “Lawyers and insurance companies are becoming increasingly involved in design decisions as liability becomes a bigger impact,” says Rupert Baines, CEO, UltraSoC. “These are unusual and sometime uncomfortable conversations. A couple of years ago, what semiconductor person would expect to have to explain their technology to a lawyer, an actuary or an underwriter? Everyone has to get their heads up and see the bigger picture.”

Autonomous driving may go through a period of slower development. “Autonomous driving deployment will be stuck at Level 2 even in 2020, says Cadence’s Wong. “There is lots of talk about Level 3, but I predict that most OEMs will continue to improve Level 2 technology and capabilities and lower the cost/price so they can put them into lower priced vehicles. They are likely to skip Level 3 altogether and move directly to Level 4, but we won’t see that until 2021 or even 2022.”

Electric vehicles development is certainly progressing rapidly. “Registration for electric cars was up 125% in 2019 and this growth will increase in 2020 as a broad range of cars become available,” adds Jones. “Significant financial incentives to move to EVs increase the likelihood of this happening while consumer concerns will reduce as battery technology continues to yield a 6% to 8% year-on-year improvement. Fast charging technology will start to be rolled out more broadly, easing another of the main issues for consumers.”



Leave a Reply


(Note: This name will be displayed publicly)