So you’ve been toiling away in the depths of the EDA world and you are struck by an idea of monumental brilliance and potential. You drop what you’re doing and go off into a cave for a while to flesh it out to the point where you can solicit a hearty investment by a forward-thinking manager of an aggressive investment fund.
What are your chances?
It certainly won’t come as a surprise that you’ve got more than one roadblock to get by. It’s not an easy investment environment out there – for high tech in general (at least for anything you can actually put your hands on). Even tougher for EDA.
So… what does that mean for EDA? Does it mean the industry will slowly wither away for lack of funding as development costs go up but revenues struggle to budge? If that happened, then the silicon industry would slowly die, and, if that happened, there would be no new platforms on which to run your precious Googles and Facebooks and all the other pretty darlings of investment these days. Is that how it all ends?
Not according to Lucio.
If you’ve spent any energy near the investment wing of the EDA world, you know who Lucio is. He’s Lucio Lanza, storied investor in EDA – someone who has stuck around while most other big guys have gone looking for a different party to crash. And, as such, he was this year’s recipient of the prestigious Phil Kaufmann Award, given out by the EDA Consortium coincident with the ICCAD conference.
This award has spawned many press stories of his life and career. Given an opportunity for a conversation with him, I figured that the n+1st such story might not be productive. So what we did talk about was his thoughts on the investment environment for EDA – and for hard tech in general. (Yeah, I just dissed all the social media stuff by suggesting it’s soft tech. Or easy tech; take your pick.)
To be clear, Lucio says that, by definition, investors are optimists. He certainly is. But, in the same breath, he also described investing in EDA as an “exercise in depression.” And that’s not only EDA – that’s also a fact for semiconductors and other geeky high-tech. ROIs just aren’t that great.
As Lucio describes the various causes for this difficulty, they can be summed up in a single word: complexity. There’s more nuance than that, but, in the end, the arrows all point to complexity.
Going for Liquid
Let’s start with silicon. Each node is harder than the prior. And that means that adoption is taking longer and longer. And that’s as planned; in reality, it takes even longer than the longer plan, as Intel has found with the 14-nm node. It also means that EDA has a harder job dealing with all of the new stuff in the new node.
The changes may not all be on the order of the new FinFET device, but they still add up. And – critically – designers won’t buy and use the new-and-improved tools on the new-and-improved process node until that node is in production.
In other words, it is taking longer and longer to reap the benefits of an investment.
An important point here: tech investors tend to like technology – especially these days, now that the money-is-the-only-motivator investors have fled. But what is it that defines the success of an investment? Is it watching the amazing new gadgets being scooped up by consumers? Gadgets that exist due to the advances in technology that your investment brought about? Pride of ownership? Nope. That pride may well be there, but it’s not the goal. The goal is no different here than it is for any other investment: liquidity. The goal is getting your investment back with a healthy return. The goal is “divesting,” if you will. You get in so that you can get out.
So a key metric in any investment is “time to liquidity.” When should I expect a “liquidity event” like a buyout or IPO?
This isn’t just the, “Are we there yet?” pesterings of an impatient investor. It turns out that there’s a very practical reason why this matters: many investment funds have a specified lifetime. If you have a ten-year fund and it takes seven years for the investment to exhibit any visible result, that’s too late. It’s not viable. And many funds do have a ten-year sunset. And seven years is the current median time-to-liquidity. You see the problem.
In fact, Lucio suggests that this is why much of the investment comes from corporate investment arms like Intel Capital. Their funds tend to be “evergreen” – they can continue to take new money, and they don’t have a specific expiration. In other words, they can afford to be more patient. In the absence of big Sand Hill firms, it’s corporate funds and small investment funds like Lucio’s that keep things moving forward.
As to the most likely form of liquidity, EDA IPOs are rare these days. Lucio thought back to the last major IPO in EDA: 2001, with either Magma or PDF Solutions. Since then, liquidity has taken the form of buyouts – typically by one of the Big 3 firms (Synopsys, Cadence, and Mentor Graphics). To some of us, it looks like companies may have given up on the IPO track. But Lucio suggests that IPO is still a goal and that many of the buyouts have happened while the acquired company was readying itself for an IPO.
He sees this as good for both the acquired company and the buyer. The big companies tend to get too locked into their ways to come up with dramatically new and improved ideas. Bringing an outside company in – and then nurturing (rather than stifling) it – can be a great way for the big guys to participate in the innovation that’s so necessary to keep from being passed by.
Complexity has another impact: how the heck is anyone supposed to understand what they’re investing in? That’s not because the investment itself has to be complex; it’s almost the opposite. New startups can’t hope to take on too much, so your typical idea will focus on one thing. It will be a narrow solution, most likely involving some deep, impenetrable concept that would elude most mortal folk.
If it takes a few PhDs to identify a problem and find a cure for it, how the heck is someone with only a finance degree, for instance, or even a tech degree 20 years ago, supposed to decide whether this investment is a good idea? Pictures of cats, on the other hand: that’s a sure winner. [OK, to be clear, that’s my snark… not Lucio’s. I didn’t find snark to be his style…]
Complexity in the System
So, with all of the impact of this complexity, what is its cause? It’s more than simply the mind-boggling minutiae of building ridiculously small features on a silicon wafer. There’s also higher-level complexity in the fact that we’re now putting entire systems on a single chip – the aptly-named system-on-chip (SoC).
You could argue that we’re not doing anything now that we weren’t doing before – all we’re doing is collecting it all up and putting it on one die rather than piecing it together and connecting it on a PC board.
And you would be right. But what might have been a custom-designed chip before is now a piece of IP. There are details that must be managed in integrating IP. And there’s a delicate balance between providing shrink-wrapped circuit solutions and providing ways for circuit designers to configure IP in a way that lets them add value. If all IP looks alike, then all chips start to look alike. Lucio sees IP that allows differentiation as having an advantage.
In fact, he sees IP generally as showing more growth than tools. Which is interesting, given that, when IP first emerged as a business, it was fated to death – and IP sales were typically a teaser to the real revenue – consulting to make the IP work in a real system. Now IP sales scale more easily, and consulting may be seen as a distraction (since it doesn’t scale). That’s incredible progress.
So EDA is bearing the burden of supporting new process nodes and new design methodologies like IP. But that’s not the worst of it. There’s an entirely new responsibility that comes with development of a complete system (or the main elements of a complete system, for those that quibble that SoCs on their own aren’t complete): software.
Software tools are a whole different thing, and they take a ton of effort. It’s also a different world, since there’s not nearly as much revenue tied up in software tools as there is in silicon tools. But software verification has now gone into the critical path of ensuring that masks can be cut for an SoC.
Lucio sees a thousand-processor chip in the future. Putting all that hardware together will be an incremental effort from what we can do today, but getting software to run on it will be a much bigger deal. And he sees some software being hardened into silicon as well. There’s opportunity here.
Finally, there’s one more area where Lucio sees promise: bringing big data to semiconductor manufacturing. There should be lots of good information buried in the millions of little pieces of manufacturing and inspection data – if you can bring them together and make sense out of them. This may be where the next big wave of silicon efficiencies comes from. Not quite EDA, although it’s entirely likely that some of what’s learned will push back into silicon design.
Looking for money?
So let’s come back to the initial scenario: you’re the one coming out of the cave with a prototype for a good idea. How are you going to get funded?
Well, let’s start with this question: Does anyone know who you are? If not – there’s no way to sugar-coat this: your chances aren’t great. It has nothing to do with your idea and everything to do with the fact that investors like to back folks that they’ve invested in before. (With success, although there is that thing about investors liking leaders that have seen failure before…)
Which means you need to hook up with someone who has a known pedigree. Yes, this really matters.
Then you’re going to need to be able to articulate the value of your idea, which most likely addresses a very narrow slice of the technology pie, in a way that is understandable to investors. And don’t go for the big firms; look for more specialized small firms and corporate funds.
And is an IPO a possibility? Lucio says yes, but as I listened to his response, it’s clear that the quants have had an impact on the overall investment world: The Calculus (i.e., Newton and Leibnitz) has become explicit in the investment calculus. They look at the rate of revenue growth and its acceleration. Or, stated more succinctly, the first and second derivatives. (Yes, more financial derivatives.) They’ve got to look good if you’re going to debut on Wall St.
He sees the profits in the leading edge, which means you can’t sit back. He particularly likes companies like ARM that plow profits back into the technology. He also suggests that SoC companies shouldn’t squeeze the EDA guys too hard on prices: let them earn some profit that they can then use for further innovation.
All of that said, Lucio notes that you’re more likely to win in Las Vegas than to succeed with a new company. That’s why the bar is set so high for what can be considered a successful liquidity event: the few winners have to more than compensate for the many ventures that don’t win.
And despite all that, Lucio remains optimistic. He sees no fundamental roadblocks to progress, just lots of slogging away needed to clear out the innumerable details, all of which have to be handled in order for things to move forward.
We need optimists like Lucio. And then we need to work our tails off to prove worthy of their optimism. And when we do, folks like Lucio will be cheering us on… and continuing to invest.