Everyone has a bright idea about how to save power or boost performance, but how do you make sure your startup lasts the distance?
Startups are often created by experienced engineers who figure out how to solve a technical problem they are dealing with at work, or by PhD candidates in research labs before they have even started their first full-time job. Either way, getting seed money to the tune of a few million dollars is relatively easy compared to securing further rounds of funding and achieving the company’s exit goal.
Primarily, reputation is everything for startups seeking funding, whether they are asking for money from individuals and companies in the semiconductor industry or from venture capitalists (VCs).
“The semiconductor industry is very large in terms of dollar value, but it’s somewhat insular in terms of everyone seems to know everyone, or is second-degree connected,” said Paul Karazuba, vice president of marketing at Expedera, currently in its Series B round of funding. “Being able to secure this money was not only due to the reputation of the company, but also the people within the company and the founders.”
A startup’s solution must be needed by semiconductor designers and architects, but it can’t be pie in the sky. “Very few companies make it to the Series B, generally because they don’t have the right product market fit or their technology is not achievable,” said Karazuba. “A lot of startups are trying to answer a question that no one is asking. For Expedera, people need lower power, high efficiency, edge AI inference, and we’ve been able to validate what we do through customer traction. That is absolutely key. Not only having a good story, but having the data and the market traction to back up that story, as well as being able to show paths to higher revenue and profitability are key to follow-on rounds.”
A typical startup approach will identify a problem, come up with a plausible implementation, and get potential paying customers on board to convince the VC money folks that it’s worth investing in. “Then it starts to snowball,” said Steve Roddy, chief marketing officer at Quadric. “We had a corporate strategic customer, Denso, anchor the Series B round, because they’re an early user. You get an early anchor that brings along venture money. When a VC sees a paying customer is willing to set the price and invest, multi-millions of dollars will follow along. We’re days away from closing our Series C funding, and it’s going to be completely financial investors, pure growth VC money. There’s still support from the previous corporate investors, but it’s being led by an independent VC firm. That’s the classic pathway. Get an idea, convince someone your idea is strong, build a proof of concept, make a sale, and then start scaling.”
Tips on the startup funding process from Karazuba and Roddy:
Each series of funding has milestones in which to demonstrate what you’ve achieved before you get to the next one.
“At each point, investors look at the technology assets, quality of the engineering and execution leadership that allows you to get from point A to point B and C and then D, but more often than not, in the early stages, they’re betting on the team,” said Nandan Nayampally, chief commercial officer at Baya Systems, currently Series B. “We saw an opportunity to go bigger and raised our Series B to accelerate the path to revenue generation. You take funding to tide you through the technology investment and development phase — the tough times as you establish yourself, gain credibility get to the next stage. We’ve moved very quickly down the revenue path. That has done two things. One, it has validated that the demand is there, and that our technology and our model are right. Two, it delays the need for the next round, because we’ve just increased our runway in our path to break even, and then to profitability. Our effective burn rate is lower, because while we’re using investor funds, there’s revenue coming in. It’s not just outlook. We’re moving well on that path, and we’re beginning to get traction in multiple markets.”
The advantage of delaying C is that at every stage, a company needs to decide what it’s trying to do with that investment. “Every time you take funding, you dilute yourself,” said Nayampally, who previously was involved with multiple VC-funded startups, a bootstrapped startup, and closely engaged with Arm’s incubation investments while working at the company for a decade and a half. “To take more funding, you’d better know why you need it. Investors will only provide more funding if the size of the market you’re going after is bigger and they believe you have the technology and team to win that and raise the valuation of the business. Initially, we were trying to prove ourselves. ‘Here’s the technology. It has a market like this.’ When we went to Series B, we expanded our scope and said, ‘Actually, the total accessible market (TAM) for us is much bigger.’ Hence, when we got the series B, we had a better valuation because the investors saw the potential of that market and our capacity to grow into it. As we prove what we’ve done with Series B, we can either expand our scope, expand product lines, identify a step-change in the size of the market, and decide, ‘To get there, we will need new investment,’ or execute on rapid growth with the current plan to profitability and expand the business more organically.”
Finding money for prototyping and failing fast
Seed funding is typically obtained through a startup’s network, but VCs are needed at the Series A, B, and C levels to provide tens of millions of dollars, at least, because hiring a team and building company infrastructure is expensive. “That includes the hardware and the software to set it up, and then it’s a very iterative process,” said Vikram Bhatia, executive director of Synopsys Cloud, which has a startup program to help get design collateral to foundries faster. “You have to fail fast. You have to get something out and test it. If it doesn’t work, go back, make changes, pivot, and change the design. Because of that, the funding requirements are very large. I have seen many startups have a prototype out but not have revenue, so they sell pre-revenue but still get to Series A, because everybody who’s active in this industry — in terms of people who are funding it — knows that getting a real design out and gaining market adoption is a long game, and requires a lot of investment.”
Incubators can also help startups in the early stages. Silicon Catalyst Angels is a separate, independently run organization from Silicon Catalyst, formed because some of the latter’s advisors were interested in investing. “It gave the startups a way of raising the relatively small sums of money they needed for seed funding,” said Nick Kepler, chief operating officer at Silicon Catalyst.
Then, about a year and a half ago, the Silicon Catalyst Venture Group was formed, also separate from Silicon Catalyst. “That was a group of people who came to us and said, ‘We want to invest in semiconductor hardware companies. We would like to work with you on identifying ones that we can invest in.’ They invested in over 10 of our companies in the last year — very, very fast. They’ve raised over $15 million, and that’s been really beneficial for them as investors, but also for our companies, and being able to get investment dollars. It allows our companies to go build their proof of concepts, to do their initial customer engagement, and then prove they are worthy of the larger investments that will come from the major VCs.”
A solid proof of concept is essential. “A startup company tends to be founded in this field by a small number of engineers who are very technical people, and they’ve got some novel idea that is transformative, and it’s going to be great, but their training doesn’t help them understand, ‘Okay, so what should I do with it?’” said Kepler. “It’s also quite common for a company to underestimate how much it’s going to cost them to get to market, to overestimate how much they’re going to be able to sell their product for, or to underestimate how much their competition is going to do.”
Another way to get capital or access to infrastructure is through a government program like ME Commons or SBIR. “I don’t think there is anything inherently better or worse about taking money from that versus a VC,” said Quadric’s Roddy. “It depends upon what you’re doing as a company, the product that you have, the situation that you’re in, and the intricacies of how the money is being given.”
That said, some startups prefer VCs to government programs. “In general, venture capital is the best way to get enough funding,” said Andy Heinig, head of the Department for Efficient Electronics at Fraunhofer IIS’ Engineering of Adaptive Systems Division. “I often see that what you can get through public funding is not that much, or not enough to go forward in a fast way. But there are some exceptions. In Europe, we see certain money available on the defense side. This is much more than before, so this may be another way to develop new ideas, under the umbrella of defense.”
Wherever the investment comes from, the biggest enemy of silicon startups is time. Sometimes the runway is a year or less to get the first prototype up. A further pressure is that design time must coincide with a foundry’s startup shuttle. “Foundries typically reserve capacity for large organizations, where the entire foundry is going to just focus on their chip for the next six months,” said Bhatia. “But if you look at a startup, they’re not going to build millions. Especially if they’re doing a prototype, maybe it’s just a few hundred or a few thousand chips, so the foundries create dedicated shuttles [for this purpose]. They say, ‘If I get your final design for tape out before this date, then you are on the shuttle. Otherwise, you miss the bus and need to take the next startup shuttle.’ Many of the startups we have worked with, especially over the last year and a half, have taped out. We know what happens. Your shuttle is going out in three months, and you’re behind. You need access to more hardware, more software, to get the verification done, to get the bugs out. It’s like the Wild West. They want to achieve success because they know if they miss the shuttle, the next one’s going to be months later.”
Prove a 10x improvement in a challenging area
Many startups are big on promises, but they must have hard evidence that they can deliver a substantial benefit to customers.
To that end, Siemens EDA has a VC fund and actively works with universities and startups to find solutions to its customers’ problems. “Fundamentally, the startup’s solution needs to be a game changer in terms of achieving a 10x improvement,” said Sathishkumar Balasubramanian, head of products at Siemens EDA. “It could be in area savings, power savings, or reducing the tape-out schedule. We call it a 10Xer. That is the value add. It captures people’s attention. Then you don’t need to go ask for funding. People are going to come and ask, ‘How can I work with you?’ You will be called into any conference and everything else. It’s really a function of how smart you are in reading the industry, the customer challenges, and thinking further ahead.”
For example, ChipAgents raised $3.09 million in its pre-seed funding round by promising a 10X improvement in productivity. Kexun Zhang, head of research at ChipAgents and PhD student at Carnegie Mellon University, emphasized the importance of talking to customers and learning from them. “I’m more on the research side of this company, which is usually the side that talks less with customers, but whenever I talk with them and hear from them about their needs and their problems, the challenges they face, I get so much information about what good research directions there are, and the problems to solve that are going to create the most value.”
While many startups have a good idea, only about 10% of those that apply to Silicon Catalyst make it past screening. “We want a group of highly technical engineers with a unique technical solution to a big problem and the intellectual property to protect it –– patents they have written around their invention, or patents they have licensed from a university they came out of,” said Kepler.
One of the reasons a startup may not pass screening is simply because their idea is not new. “Our advisors can tell a company, ‘I’m sorry, but what you’re doing, several other people have done in the past, and they tried to make businesses out of it, and here’s what’s gone wrong, and here’s why you’re going to fail,'” said Kepler.
Challenges in need of solutions
The semiconductor industry has a growing list of challenges that need fresh innovation and funding to enable future technology.
“Funding often follows vision,” noted Steven Woo, fellow and distinguished inventor at Rambus. “Companies and students should focus on solving real-world bottlenecks with clear technical differentiation. If resources were unlimited, we would accelerate development of power-efficient, high-performance, low-latency memory systems purpose-built for AI at scale. Future systems will also need to focus on increasing security and reliability, and accelerating technology and product development in these areas would be a focus.”
Chip designers are also in need of solutions for 3D-IC multiphysics effects, electromigration and IR drop (EMIR) coverage, and analog design. “Analog design has been rising in importance because frequencies are higher,” said Marc Swinnen, director of product marketing at Ansys, now part of Synopsys. “There’s a lot more video and interfacing, which is more analog than the old digital screens. It’s always been there. It’s always been important, but now it’s rising.”
Nilesh Kamdar, general manager for the design and verification business unit at Keysight Technologies, agreed that high-frequency and high-speed technology is one of the big-ticket technology changes that are coming, along with AI. “Doing design using AI techniques is a huge investment and requires many more ideas, some of which are going to fail. Some of them are not going to give us the benefits that we seek, but then a few of them are going to be really impactful. Chiplets and all forms of advanced packaging also need a lot more investment. It’s not that there hasn’t been any, but there can be much more.”
On-device AI is another massive growth area. “AI workloads are growing and evolving faster than ever, but so are user expectations around battery life, thermal efficiency, and real-time responsiveness,” said Chris Bergey, senior vice president and general manager of Arm’s client line of business. “We’ve focused on the essential requirements for on-device AI, namely high performance for real-time inferencing at low latency to avoid unnecessary round-trips to the cloud, and energy efficiency that doesn’t compromise capability or usage.”
Finally, more verification solutions are needed. “With shift left, we’re not trying to eliminate post-silicon, but as the chips get bigger, the workloads get more complex, there’s so much that has to get done post-silicon that you can’t help but pull that into the pre-silicon process as much as possible, simply because there’s no time. You want to parallelize it as much as possible, and it’s easier to debug and fix before you’ve built the actual device,” said Matthew Graham, senior group director, verification software product management at Cadence. “Our customers are demanding that they pull more and more into the pre-silicon phase, for time to market, for quality, for cost.”
Conclusion
Overall, the chances of success for a silicon company are higher than for a typical software or technology startup. “One of the reasons is that silicon design is unique. There’s a very niche market of talent that goes after this, and they only go after this because they have skills and expertise,” observed Synopsys’ Bhatia. “They are helped by the support from ecosystem companies like us, incubators, and some large semiconductor design companies that have their own venture funds. We’ve got a profile of a founder who’s been in the industry for 15, 20 years, and now he’s building a startup. He has credibility, he has a network, so the success rate is much higher.”
In terms of exit goals, going public is rare. “We view it as being a perfectly good outcome to be acquired by a company that values what you’re doing enough to purchase you, and then to be able to join that company and continue within that company, to drive your vision forward as a product with the heft of that larger company behind you,” Kepler said.
Remaining independent is tricky because of the quandary of how to repay the investors. “There are companies that do that, and have that goal,” he continued. “They use their own money, then they use their first revenue and bootstrap themselves into a position where they can form a business and become profitable. It’s harder to do than if you have investors, because you have to find the money or do things on a shoestring. If you’ve brought in external investors, they’re going to expect you to have an exit that brings them a return.”
Related Reading
Government Funding For Chip Design Tools Spreads
Manufacturing has the lion’s share of global government handouts, but design R&D is gaining ground as the ecosystem draws closer.
Navigating The Challenges Of Group Design Projects
As more companies and startups join forces with government and academia in chip design projects, issues around data sharing, IP protection, and PDK access must be solved.
 |
 |
 |
 |
 |  |
 |  |
 |
Leave a Reply