Never miss a story from Breakfast Bytes. Subscribe for in-depth analysis and articles.
A couple of weeks ago, Silicon Catalyst organized an evening panel about corporate venture capital. Corporate VCs are groups within companies, perhaps Intel Capital is the most well-known, that make investments as opposed to the sort of VCs all along Sand Hill Road, which are known as financial investors. Corporate VCs vary in what their goals are, ranging from simply trying to get a better return on the company's cash, to investing in companies that they might acquire if all goes well, and everything in between.
First a word about Silicon Catalyst. They are an incubator focused entirely on semiconductor startups. They don't make investments in the companies themselves, they provide facilities and negotiate deals on behalf of their entire portfolio for cheap or free design tools, shuttle runs, and so on. They also help their startup companies raise money. Since financial VCs are not interested in semiconductor (they are busy chasing the next Uber for X), this means corporate VCs such as those on the panel at Wilson Sonsini's HQ in Palo Alto that evening.
What sort of investments do you make?
Applied: Semiconductor and display technology. Made 65 investments to date. Some in robotics, DNA sequencing, healthcare.
Intel: Everything. They have made 1400 investments over the years, of which about 600 have been acquired or gone public. Investments vary from seed capital all the way up to $3B in ASML (for EUV). Currently very interested in deep learning and technologies that complement FPGAs post-Altera.
Qualcomm: Anything that puts a lot of computation at the edges of the network. Computer vision especially interesting right now. Sometimes $250K. Largest is $25M. "We've not done any $3B deals!"
SanDisk: Mostly do series B investments since they are looking for users of flash and so they like to see something working. Most investments $3Mish.
Cypress: Only do strategic investments when Cypress can help with assets or market access. They do not do purely financial investments. Mostly B round. Their most famous investment is SunPower which became larger than Cypress itself.
What has changed since you started?
Cypress: Semiconductor is a difficult field from an investment RoI point of view. Even so, everything has more investment in software than hardware these days.
SanDisk: Now there are thousands of companies investing. But the model has changed. It used to be a financial VC leading who wanted a corporate logo to put on the slide-deck for their roadshow, but now corporate VCs often lead the round.
Qualcomm: Analog is easier since it is no longer feasible to fund a traditional large digital SoC. There are now very few financial VCs, typically only Walden (Lip-Bu) have the background. There are lots of semiconductor startups in China these days, which is new. But there is huge consolidation of the ecosystem with only three companies now funding non-memory fabs and only three plausible exits for an EDA startup.
Intel: There is more consolidation to come in areas that haven't yet consolidated (e.g., RF, MEMS). There has been a change in why people come to corporate VCs, since financial VCs can only offer limited help, whereas a company like Intel can offer access to customers, technology co-development, weight in stardards development. It used to be that the investor would ask what the startup can do for them. That has now flipped and the startup is asking what the investor can do.
Applied: Now more innovation centers, not just Silicon Valley. The process is better understood so it is no longer shooting in the dark about how to navigate the system.
How do you balance RoI versus strategic goals?
Qualcomm: Purely financial goals, with no need for business unit buy-in. Need to make good financial investments or the treasurer will turn off the money tap. But strategic is hard since "corporate strategy changes very year".
Intel: More an art than a science. If the investment is important strategically then might relax the financial goals and vice versa.
Applied: Need the possibility of a strong exit or else it is hard to see how it can be strategically succcessful.
SanDisk: CFO does sit on the committee, and they do have financial metrics. But mostly they invest in the flash ecosystem.
Cypress: Mostly strategic and usually involves a business unit, too.
What is state of semiconductor startup ecosystem?
Applied: Quantity not high but quality very high.
Intel: Lots of opportunity and less competition. Financial VCs not going into the space so pricing reasonable unlike in consumer internet or social media.
Qualcomm: You need a 5-10-year horizon. No longer just Silicon Valley though, China and Israel and Nordic countries (for power and analog).
SanDisk: There is a trend towards building full devices, not just a chip or IP. Smart watch, not a smart watch SoC.
What are your decision criteria?
Cypress: Plan A never works so need breadth of skillset in the team, and flexibility.
SanDisk: "It is a fine line between confidence and delusion." Need team that can adapt.
Qualcomm: Why you? Why this problem? Why now? Too early, no market. Too late is better, at least you get a piece. "Being introduced by someone I really trust is important."
Intel: We do a postmortem every quarter on what succeeded, what was a complete failure. Management is key. Market opportunity (high growth) is the other. Engineering not completing the product is rarely the issue.
Applied: What is the high-value problem that you are solving? If competing against established players, is the differentiation sustainable for five to eight years or can they produce a good enough product, too?
Audience: Do you ask for conditional exits?
(What this means is having verbiage in the investment documents that gives right of first refusal in an acquisition.)
Qualcomm: Only do financial-style deals so "no".
SanDisk: No "try before you buy". Just want notification rights (they get told of an acquisition so can maybe counter).
Intel: Sometimes ask for notification. Corporate VC is more mature now and no need to try and capture everything.
Cypress: "Putting handcuffs to force the outcome can damage the outcome."
Audience: Do you find the companies or do they find you?
Everyone had pretty much the same answer that it was a mixture. There is lots of stuff that comes in from Sand Hill Road when they aren't interested but like the look of the market/team/idea. But they do lots of outbound stuff. "That's why we are all here tonight."
Cypress: Keep trying, it can be done.
Intel: Yes, there is a myth of the overnight success. "Mostly it is a ten-year overnight success." Fail fast and pivot.
SanDisk: I would totally echo that and I don't know why people say VCs have a herd mentality!
For the Silicon Catalyst website, click on their logo to the right.
Previous: CDNLive: Ericsson Paving the Way for 5G
Next: CDNLive: Design Technology Co-Optimization for N7 and N5