Intel has announced its intention to acquire Movidius. My firm, Robert Bosch Venture Capital (RBVC), invested in Movidius in 2013 alongside other investors, including Atlantic Bridge and DFJ Esprit. We are proud to be a part of the success story of Movidius, and congratulate the CEO Remi El-Ouazzane, the founders Sean Mitchell, David Moloney, Valentin Muresan and the entire team for a great achievement. I wish them all the best and much success as they start the next chapter of Movidius as part of the Intel family.
We faced many concerns and doubts when we started our due diligence of Movidius at end of 2012. Just one example – people questioned whether the world needed another image signal processing (ISP) company. This discussion may look strange today, but that was the perception at that time on Movidius’s technology, because the only visible application for Movidius’s processor then was computational image processing, such as array cameras, for mobile devices where the chip would act as a co-processor and an enhanced ISP. Another concern we often heard was the general risk of investing in a semiconductor processor company – high capital requirements, long development cycles and so on.
In the end we decided to invest in Movidius because we believed in the team and the disruption potential of their concept of a low-power vision processing unit (VPU) in multiple industrial segments. It was quite a risk given that at the time the “killer application” was not yet foreseeable. Nevertheless, with the industrial insights our mother company Bosch has, we were quite optimistic about the potential of this technology in the areas of security, automotive, power tools and augmented reality.
The breathtaking rapid development of Movidius in the past three years is history. In 2014 Google announced its Tango device, for which Movidius was a key partner providing the image processing platform. The autonomous drone engagement with DJI in 2015, the engagements with Flir in their smart thermal camera, and with Lenovo in the virtual and augmented reality area are only a few of the highlights in a series of significant design wins by Movidius. Another one is the recent announcement by Google of a broad collaboration with Movidius in embedded deep learning. It seems to be obvious that Movidius has become the place to go for low-power vision intelligence applications.
As an investor, RBVC supported Movidius in multiple ways beyond the capital provided to the company. For example, we help with domain expertise, including technology benchmarking, insights in markets, and applications in the relevant industry segments. RBVC also tried to help the company in building a rich ecosystem, and some of RBVC’s portfolio companies are working very closely with Movidius.
From the technology perspective, Movidius’s VPU is enabling a huge amount of different features, such as camera imaging pipelines, 3D point cloud generation, simultaneous localisation and mapping, pattern recognition, object tracking and other machine learning tasks in a very efficient way. All you need to do is to use the small, low-cost and ultra-low-power chip Myriad 2. The combination of massive computational power for vision and the low-power battery-driven device enables real embedded machine vision applications for the first time.
Today we see that artificial intelligence (AI) is becoming ubiquitous. Such intelligence is going to move from the cloud towards the edge – to mobile and portable devices. Obviously Movidius owns a key enabling technology supporting this trend for multiple industry segments in the decade to come. I am confident that Movidius’s VPU technology will go way beyond the applications the world has been excited about in the past three years.
I will be following the journey of the Movidius team in the future under Intel’s umbrella with great interest. I wholeheartedly wish them the best of success in the years to come and once again would like to thank the whole team for their relentless effort and commitment in the past years.
Let me finish with a point of view on the other concern we faced as a semiconductor VC investor. Can the success of Movidius be repeated in today’s environment? My personal answer as an investor is yes. I believe that the time for special semiconductor investments is just beginning. We have seen much excitement recently in the cloud-based deep learning accelerator area. There will still be opportunities in this area and I expect further huge movement in the coming months and years. Semiconductor technology is the backbone of artificial intelligence. As the world is moving rapidly towards the AI age, huge opportunities for hardware startups are emerging.