What is the state of innovation in embedded vision? While deep learning remains a dominant force, deep neural networks alone don't make a product. Presented as a virtual event in May, the Embedded Vision Summit examined the latest developments in practical computer vision and AI edge processing. In my role as the summit's general chair, I reviewed more than 300 great session proposals for the conference. Here are the trends I'm seeing in the embedded-vision space.

A staggering amount of money is pouring into data center AI chip companies at the moment. Data center AI chip companies are raising eye-watering amounts of money. In the last week, we've seen Groq announce a $300 million Series C round of funding, and SambaNova raise a staggering $676 million Series D. SambaNova is now valued at somewhere above $5 billion. They are not the only ones in this sector raising these huge amounts of money. Fellow data center AI chip companies Graphcore (raised $710 million, valued at $2.77 billion) and Cerebras (raised more than $475 million, valued at $2.4 billion) are hot on their heels as the sector continues to gain momentum.

Since the novel coronavirus (COVID-19) outbreak in early 2020, businesses across multiple vertical industries around the world, have struggled to deal with unprecedented challenges posed by the pandemic. With millions of workers furloughed and companies shuttered, the virus has had an enormous impact in both human and economic terms. While many industries have suffered, opportunities have also been created, especially in B2B and the Internet of Things (IoT). A new report from Strategy Analytics, "The Impact of COVID-19 on Enterprise and the IoT," identifies how the coronavirus pandemic has impacted industries of all sizes across a multitude of vertical markets, identifying key challenges as businesses reopen, as well as opportunities in the IoT, especially areas such as Telehealth, Automation, Spatial Computing, Digital Twins, Supply Chains, UAVs and Robotics. This report takes a look at the impacts caused by COVID-19 on the enterprise and IoT space and the opportunities and challenges posed as the global economy attempts to recover on the path to 2021. "The COVID-19 pandemic has created a new environment for citizens, companies and governments.

Imec and GlobalFoundries have demonstrated a processor-in-memory chip that can achieve energy efficiency up to 2900 TOPS/W, approximately two orders of magnitude above today's commercial processor-in-memory chips. The chip uses an established idea, analog computing, implemented in SRAM in GlobalFoundries' 22nm fully-depleted silicon-on-insulator (FD-SOI) process technology. Imec's analog in-memory compute (AiMC) will be available to GlobalFoundries customers as a feature that can be implemented on the company's 22FDX platform. Analog compute Analog compute, or processor-inmemory, is an established technique that is already used in commercial AI accelerator chips from startups Mythic, Syntiant, Gyrfalcon and others. Since a neural network model may have tens or hundreds of millions of weights, sending data back and forth between the memory and the processor is inefficient.

Today's smart-lighting systems still must be set up manually by the user. The OpenLicht project has developed a prototype for a more intelligent lighting system... German research project OpenLicht has successfully developed a smart-lighting system based on open-source software and machine-learning libraries, plus inexpensive hardware, that can automatically adjust lighting in a room based on what the user is doing. Today's smart-lighting solutions are based on smart bulbs such as the Philips Hue and Osram Lightify. While they offer some smart features, they generally require manual control by the user via a smartphone app. Some can be programmed (for example, to turn on and off at certain times), but the rules still have to be set up manually, so the basic relationship between user and lighting system is not changed by making it smarter.

AI chip startups are hot on the heels of GPU leader Nvidia. At the same time, there is also significant competition in data center inference... New computing models such as machine learning and quantum are becoming more important for delivering cloud services. The most immediate computing change has been the rapid adoption of ML/AI for consumer and business applications. This new model requires the processing vast amounts of data to developing usable information, and eventually building knowledge models. These models are rapidly growing in complexity – doubling every 3.5 months.

Edge computing provides an opportunity to turn AI data into real-time value across almost every industry. The intelligent edge is the next stage in the evolution and success of AI technology... As adoption rates rise for artificial intelligence and machine learning (ML), the ability to process large amounts of data in the form of algorithms for computational purposes becomes increasingly important. To help make the expanding use of data applications across billions of connected devices more efficient and valuable, there is growing momentum to migrate the processing from centralized third-party cloud servers to decentralized and localized processing on-device, commonly referred to as edge computing. According to SAR Insight & Consulting's latest AI/ML embedded chips database, the global number of AI-enabled devices with edge computing will grow at a compound annual growth rate of 64.2% during the 2019–2024 period.

Intel is sponsoring an online course to help address the shortage of AIoT developers... Amid rapid growth in AI deployments across a variety of industry sectors, Intel has decided to address the skills shortage in AI-savvy developers by partnering with online technology learning platform Udacity to offer a course in edge AI for developers. "Historically, students have learned how to build and deploy deep learning models for the cloud. With Udacity, we are training AI developers to go where the data is generated in the physical world: the edge," said Jonathan Ballon, Intel vice president and general manager, Internet of Things Group. "Optimizing direct deployment of models on edge devices requires knowledge of unique constraints like power, network bandwidth and latency, varying compute architectures and more. The skills this course delivers will allow developers -- and companies that hire them, to implement learnings on real-world applications across a variety of fields."

This market is absolutely teeming with chip startups, many of whom are reaching a level of maturity where they are revealing their architectures and starting to produce measurable results. As established semiconductor companies start to appreciate the importance of the AI accelerators, and the range of vertical markets AI will encroach on, will some of them look to jump-start their strategies with acquisitions? With dozens of startups at the stage where first products are being marketed and results are being unveiled, the opposite effect also applies. I spoke with Geoff Tate, CEO of Flex Logix, recently and he quoted Warren Buffett: "When the tide goes out, you can see who's been swimming naked." Not all the startups we see in the market today will be successful.

Inphi Corp., is buying most of eSilicon; while Synopsys will acquire the fabless vendor's embedded memory and interface intellectual property (IP) business. Inphi is to pay $216 million for eSilicon in both cash and assumption of debt, while the price that Synopsys paid for the memory assets was not disclosed. Targeting high-bandwidth networking, high-performance computing, artificial intelligence (AI) and 5G infrastructure markets, its IP includes configurable 7nm 56G/112G SerDes plus networking-optimized 16/14/7nm FinFET IP platforms featuring HBM2 PHY, ternary content-addressable memory (TCAM), specialized memory compilers and I/O libraries. Its neuASIC platform provides AI-specific IP and a modular design methodology to create ASICs. Speaking of the acquisitions, Jack Harding, president and CEO of eSilicon, said, "Our engineering talent, IP and customer relationships in networking, data-center and cloud, telecom 5G infrastructure and AI will help enhance their respective offerings." The Inphi acquisition of eSilicon is expected to close this quarter subject to US and Vietnamese regulatory approval (eSilicon has been in Vietnam since 2010).