December 11, 2019 –
As we go into 2020, my earlier prediction about Silicon Valley undergoing a renaissance in silicon seems to be coming true. Many world-class entrepreneurs are forming new companies including Fungible and NUVIA from our own portfolio. But the entrepreneurs ushering in this new golden age face many challenges. Among them: consolidation in the silicon foundry business, market and geopolitical threats from China, and the need to create silicon architectures that remain viable as software rapidly changes.
These issues were among the topics covered by a Churchill Club panel discussion that we convened with Diane Bryant, former senior executive of Intel and Google Cloud, moderating a conversation with John Hennessy, Stanford President Emeritus, Alphabet Inc. chair, and co-founder of MIPS; Pradeep Sindhu, founder of Juniper Networks and co-founder and CEO of Fungible; and Rodrigo Liang, co-founder and CEO of SambaNova Systems.
My co-panelists and I all agreed that new silicon architectures are needed to fuel innovations in AI, Cloud, IoT, and 5G. The path to that growth is both exciting and fraught. Not all the solutions will be invented by startups, as some problems require leadership from government or established companies, and some feel intractable.
Here are three of the big questions we tackled at the event.
Can Silicon Valley survive without U.S.-based silicon manufacturing?
There’s been tremendous consolidation in silicon manufacturing. The industry is now dominated by a few large foundries, including Global Foundries, Samsung and Taiwan-based TSMC. Silicon Valley remains vulnerable to things like trade wars, market availability, and geopolitical maneuvering. If you are in the semiconductor industry you need to be thinking about China and the impact it could have on Taiwan. I personally think it’s time for the US Government or a big tech company to invest in a U.S.-based silicon foundry.
What about China’s plan to become tech independent?
China has a five-year plan to become a net exporter of technology. Should U.S. tech companies, especially ones that do business in China, start to worry? China is the ultimate conundrum. It’s a large market that U.S. companies need access to, but a major competitor. A paramount concern there is intellectual-property theft.
One solution, signing joint-venture deals with the Chinese companies, is a double-edged sword. It can help a U.S. startup gain traction in an important market. But it might come at the expense of creating a potential competitor in the future.
Application specific compute in a changing software world
A number of startups are developing silicon products targeted at a given workload for things like AI and data-centric computing. Companies like SambaNova Systems and Fungible are examples of these. One challenge they face is that they’re developing a silicon product targeted at workloads like AI that are developing a rapid pace. With a three-year development time, how do these companies ensure future relevance?
I believe that the only way to do this is by having a programmable platform and creating an ecosystem where software developers builds applications on top of your silicon products.
As IoT edge devices continue to proliferate, data centers get distributed, and the rise of edge use cases such as autonomous driving go mainstream, I believe that there will be a lot of innovation in the semiconductor industry. I am looking forward to partnering with entrepreneurs who are putting silicon back into Silicon Valley.
Originally published on LinkedIn.