In a recent article we talked about how NVIDIA (NASDAQ:NVDA) seems to be dominating “artificial intelligence (AI) hardware” with their GPUs. The only other major provider of GPUs, AMD (NASDAQ:AMD), doesn’t seem to be all that interested in aggressively marketing to an AI audience. In that same article, we also talked about how Intel (NASDAQ:INTC) and Microsoft (NASDAQ:MSFT) are placing bets on FPGAs as being the future of AI chips. With somewhere between 1,500 – 2,000 AI startups out there right now, there is plenty of opportunity for more than one winner especially when you consider how big the market for AI chips is expected to get.Tractica - deep learning chip revenue to go from $513 million in 2016 to $12.2 billion by 2025 Click To Tweet
After our recent analysis of AMD GPUs vs NVIDIA GPUs, we wanted to take an inventory of all AI hardware startups to keep an eye on. The last time we did this was about a year ago when we published an article on “5 Startups Building Artificial Intelligence Chips“. Since then, one of those startups, Nirvana, was acquired by Intel. Let’s see what’s been happening with the rest of them and also take a look at a few new names we came across.
Since KnuEdge “emerged from stealth” last year, the company has gone quiet and not offered up any additional information about what they’ve been up to. According to an article in VentureBeat, we know that KnuEdge has already been generating revenue and that they were considering raising more funding this year in addition to the $100 million in “patient money” they have already raised. Their website contains next to no information aside from employee profiles. At an Xconomy conference a few weeks ago, the Company talked about “cloud-based machine intelligence as a service” that is “supposed to be rolled out sometime this year“.
Chinese startup Horizon Robotics joined forces with Intel to showcase an Advanced Driver Assistance System (ADAS) at CES 2017 which is a single camera that can see stuff and identify it. They also teamed up with Midea and launched an intelligent air conditioner that no doubt solves loads of first world problems including home security. No news yet on when we can expect their Brain Processing Unit (BPU) to debut but it was supposed to be “in early 2017“. The founder and CEO of Horizon Robotics, Dr. Kai YU, used to be the Head of Baidu Institute of Deep Learning so we’re expecting some great things from this startup which has taken in an undisclosed amount of funding so far.
This crowdfunding effort made its way into our original article on AI chips so it’s only fair we include it in this one. While it is not actually possible to pick a worse name for your startup than “krtkl”, at least the product name is manageable. Snickerdoodle is “reconfigurable hardware for building intelligent systems” (think Raspberry Pi). A crowdfunding effort for Snickerdoodle raised $224,876 and they’re currenty shipping. If you pre-order one, they’ll deliver it by summer. The palm-sized unit uses the Zynq “System on Chip” (SoC) from Xilinix (NASDAQ:XLNX).
Eyeriss is a project being worked on by MIT which is an “energy-efficient reconfigurable accelerator for deep convolutional neural networks” and it remains a project still. As investors, we start showing interest once these cool ideas have been vetted by venture capitalists who then give their stamp of approval in the form of an investment. While the technology is getting favorable press mentions, we’re waiting to see if it actually makes it out of the lab and into the hands of some savvy business people who can then monetize it.
Now let’s move on to all the additional startups we found out there that are building new AI chips or AI hardware.
Founded in 2016, Bristol startup Graphcore has taken in $30 million in funding to develop hardware that accelerates machine learning applications. Lead investors in Graphcore include Samsung and Bosch. The simple diagram seen below best describes the need for a new Intelligent Processing Unit (IPU) that is designed specifically for machine learning.
Graphcore’s IPU-Appliance for the cloud is coming in 2017 and is expected to increase both training and inference by between 10X to 100X compared to the fastest systems in use today.
Founded in 2010, Campbell California startup Wave Computing has taken in $45 million in funding, and is currently in the process of closing a $35 million Series D round, to develop a “new compute appliance that leverages the company’s patented dataflow architecture, which eliminates the need for a central processing unit (CPU) or graphic processing unit (GPU), removing the typical performance and scalability bottlenecks found with traditional deep learning solutions“.
That box seen above is said to deliver up to 1000x the performance for neural network training. You can apply for the “early access program” if you’re interested in getting your hands on this technology. General availability of Wave’s Deep Learning Computers will be in 2017.
Founded in 2016, Toronto-based startup Tenstorrent took in an undisclosed amount of seed funding this month to create “the next generation of high-performance processor ASICs, specifically engineered for deep learning and smart hardware“. Their team is made up of alumni from hardware industry leaders like NVIDIA and AMD. Not much else is known about this stealthy startup.
Founded in 2016, San Francisco startup Cerebras has taken in $25 million from a single Series A funding round. Little is known about this stealthy hardware startup. Their website says they are “a stealth mode startup backed by premier venture capitalists and the industry’s most successful technologists“. They also like to work hard but still maintain a sense of humor. That’s about all we know right now.
Founded in 2012, Texas-based startup Mythic (formerly known as Isocline) has taken in $9.5 million in funding with Draper Fisher Jurvetson as the lead investor. Prior to receiving any funding, the startup has taken in $2.5 million in grants. Mythic is developing an AI chip that “puts desktop GPU compute capabilities and deep neural networks onto a button-sized chip – with 50x higher battery life and far more data processing capabilities than competitors“. Essentially, that means you can give voice control and computer vision to any device locally without needing cloud connectivity.
Founded in 2010, Eldorado Hills, California startup ThinCI has taken in an undisclosed amount of funding to develop a technology that will bring vision processing to all devices. The ability for smart devices to have functionality like computer vision that doesn’t require regular communication to the cloud is referred to as “edge computing” or “fog computing”. That’s where ThinCI wants to play.
Founded in 2014, Newark, California startup Koniku has taken in $1.65 million in funding so far to become “the world’s first neurocomputation company“. The idea is that since the brain is the most powerful computer ever devised, why not reverse engineer it? Simple, right? Koniku is actually integrating biological neurons onto chips and has made enough progress that they claim to have AstraZeneca as a customer. Boeing has also signed on with a letter of intent to use the technology in chemical-detecting drones.
Founded in 2008, Massachusetts startup Adapteva is a fabless semiconductor company (in other words they outsource manufacturing) focusing on low power multi-core microprocessor designs. Adapteva has taken in $5.1 million in funding from investors that include mobile giant Ericsson. Their architecture called Epiphany holds a 10-25x energy efficiency advantage over traditional CPU architectures and has now shipped to over 10,000 developers. Here’s how their latest Epiphany-V chip stacks up to the competition:
If you want to dig into all the details, this paper should help with that. Possible applications include deep learning, self-driving cars, and autonomous drones.
Knowm is actually setup as a .ORG but they appear to be pursuing a for-profit enterprise. The New Mexcio startup has taken in an undisclosed amount of seed funding so far to develop a new computational framework called AHaH Computing (Anti-Hebbian and Hebbian). The gory details can be found in this publication, but the short story is that this technology aims to reduce the size and power consumption of intelligent machine learning applications by up to 9 orders of magnitude.
So there you have 12 startups (plus one project) that are trying to reinvent AI hardware and compete with the likes of NVIDIA with their GPUs and Microsoft/Intel with their FPGAs. Just as we were wrapping up this article, news broke that Google has announced the availability of their AI chip called TPU 2.0. Here’s a picture of a “TPU pod”:
Don’t stand too close to that thing because it might start reading your mind. As we talked about before, Intel and Microsoft are looking at FPGAs as the future of AI hardware so we’ll take a look in a future article how investors might be able to play that with Xilinix, the world’s leading producer of FPGAs. Intel has acquired multiple companies in the AI chip space so far including Movidius, Nervana, and MobilEye. Let’s also not forget IBM and TrueNorth. There doesn’t have to be one winner but investors who are looking for picks and shovels plays on AI like NVIDIA need to keep tabs on what everyone else is getting up to in the fast moving world of AI. Subscribe to our newsletter and we’ll do that for you.
If you pay more than $4.95 a trade, you're paying too much. Ally Invest is one of the lowest-fee brokers around so you spend less money on transaction fees and more on stocks. With more than 30 trades a quarter it drops even lower to $3.95 a trade. Open an account and begin trading today.