Who Wins ARPA-E Grants?

tl;dr: You can’t cheat your way to an ARPA-E grant by gaming the system. The program appears to be run in a remarkably unbiased fashion.

Most folks in energy start-up land know that ARPA-E – the U.S. government agency launched in 2009 to fund high-risk, high-reward energy research – has an open solicitation out for $150 million worth of new grants. So as proposal-writing fingers fly over keyboards, I thought it would be a good time to look at the projects that have won ARPA-E grants to date. (Full disclosure: Two of the nine active companies in our energy portfolio at Venrock have been touched by ARPA-E – Phononic Devices won a grant in the first funding round, and OnChip Power was named as a subcontractor on a power electronics grant.)

ARPA-E works through programs: Solicitations go out on a particular topic, like electric vehicle batteries or efficient air conditioning, and grants on that topic are awarded and managed in one big group. (Occasionally there will be an “open” solicitation that accepts applications on any topic, like the one that’s out now.) Grant winners may be a single organization (like “GE Research”) or, more commonly, a consortium with a single prime contractor and one or more subcontractors (like “Applied Materials partnered with A123Systems and Lawrence Berkeley National Lab”). Generally, multiple programs get awarded together as part of a broader funding event.

If you look at ARPA-E’s web site and add up all the grants that it’s ever announced, you’ll find that $516 million has been awarded to 183 projects across 12 programs during four funding events. (I didn’t get this data from ARPA-E – it’s culled from the agency’s public announcements – so I imagine its internal figures may be slightly different). The vital statistics are below (note that there’s one oddball in the mix – a group of six projects that were awarded in October 2010 separate from any particular solicitation – which I lumped into the “open” category):

So is there a trick to winning an ARPA-E grant? Are you favored if you’re in a certain type of organization, or from a preferred state, or part of some gerrymandered consortium? As far as I can tell, the answer is no. When I study the data, I find that:

Universities win more grants than any other type of prime contractor, but that’s no surprise. Universities have helmed 44% of total winning ARPA-E projects, followed by small companies at 30%. This doesn’t, however, indicate that you need to be at a university or start-up to win a grant – just that these are the places where high-tech innovation tends to arise. Differences between the programs prove this out: For example, lots of start-ups have been funded to pursue grid-scale energy storage and many big companies like ABB and General Atomics have R&D projects in the domain, so it’s unsurprising that the GRIDS program was dominated by these entities. In contrast, the nascent art of coaxing microbes to turn electricity into fuels is practiced almost exclusively at universities and national labs – which, unsurprisingly, claimed more than nine out of ten electrofuels grants.

Venture-backed start-ups get no special treatment. The Solyndra debacle created a perception that the Department of Energy favors VC-backed start-up companies rich in political connections. If that’s true, somebody forgot to tell ARPA-E: Projects led by VC-backed start-ups account for only 14% of ARPA-E grants totaling 19% of funds. Further, small companies that lack venture backing have won slightly more grants than those that have it (15% vs. 14%).

Finally, for what it’s worth, I can’t see any relationship between a VC firm’s level of political involvement and its portfolio’s grant-getting success. If that were the case, start-ups backed by the likes of Mohr Davidow Ventures and VantagePoint would be overrepresented among ARPA-E awardees. They’re not.

There’s no “sweet-spot” grant size to shoot for. The average ARPA-E grant has awarded $2.8 million, with a range of $397k to $9.2 million (the latter won by laser drilling specialist Foro Energy). But the range is all over the map. As you can see below, it’s not a normal distribution: There’s a big peak for projects in the $500k to $1 million range, but then another for those from $2.5 million to $3 million, and lots of scatter throughout. Better to ask for the amount of money it’s really going to take than try to scale your project toward some optimal target amount.

With that said, grant values have varied quite a bit by program. The air conditioning program BEET-IT exhibits the smallest average grant size of $1.8 million, while open solicitation projects claim the biggest at $3.7 million.

While geographic distribution mirrors tech hubs, 34 states are represented. Only two states claim a double-digit percentage of prime contractors on ARPA-E grants: California and Massachusetts, with 22% and 13% respectively. This should shock nobody, as these two states are home to the country’s two biggest technology clusters. New York snags 7%, Ohio and Texas each have 5%, and 29 additional states are present with a smaller share (i.e., somewhere between one and nine grants). Long story short, a little more than two-thirds of the U.S. is represented among ARPA-E prime contractors, including states like Alabama and Kentucky that aren’t generally known for tech entrepreneurship.

I wanted to see if any states were over- or underrepresented, so I compared the percentage of ARPA-E grants won in each state with the percentage of science and engineering PhDs granted there (using data from the NSF). States with a greater share of ARPA-E grants than PhDs can be considered overrepresented in the program and vice versa. Of the top five grant winners, CA and MA are greatly overrepresented (by +54% and +109%, respectively), whereas Ohio is slightly overrepresented (+28%) and New York and Texas are somewhat underrepresented (-28% and -32%, respectively). I attribute this to the fact that the overrepresented states tend to have robust research and commercialization initiatives around clean energy – the MIT Energy Initiative is a prime example – while such programs seem less prevalent in the underrepresented states.

Piling on more subcontractors won’t help you win a grant. Conventional wisdom holds that Boeing and Lockheed win military deals by subcontracting the work across partners in many states, so that lots of different stakeholders have a vested interest in the winner. This strategy does not appear to work at ARPA-E. The lion’s share of projects have between zero and two subcontractors; only 20% pile on three subcontractors or more. There are substantial program-to-program variations in this regard – the ADEPT power electronics program has an average of 2.7 subs per project, presumably because it requires multiple specialists in the electronics value chain to work together, whereas every winning project in the PETRO energy crops program was a solo entry – but nowhere do large consortia dominate.

So there you have it: ARPA-E is an extraordinary funding resource for energy technology developers, and it appears highly resistant to being gamed. Best of luck to the applicants for this open solicitation – and if you’d like to have an honest and fact-driven discussion about commercializing your technology, feel free to drop me a line.

Posted in Current events, Numbers | 6 Comments

Green Button Initiative: Great Start, More to Do

tl;dr: Data about energy consumption = good! Data about rate plans plus meaningful incentives = even better!

I know, I know, I’ve been in absentia from this blog lately: Blame the month of January, in which entrepreneurs emerge from hiding to pitch investors all at once, like cicadas from the earth. As slammed as I am, here’s a quick word on an encouraging smart grid development from a couple weeks back.

At a California event on January 18, U.S. CTO Aneesh Chopra unveiled the first implementations of the Green Button initiative, in which utilities make smart meter data available online for customers via an eponymous button on their web sites. Utility behemoths PG&E and SDG&E debuted the scheme, sharing data in spreadsheet form (for humans) and via the recently-ratified Energy Services Provider Interface standard (for third-party apps). Right on cue, software developers including Tendril and Simple Energy showed how they could import Green Button data into their apps, like energy dashboards and competitions.

Having bemoaned the smart grid’s comedy of errors in the past, I think this could mark a turning point where utilities have been successfully pressured into customer-friendliness. Utilities have no reputation for transparency, but in the wake of billions of dollars in taxpayer-subsidized smart meter rollouts, they’re under pressure to show regulators and ratepayers that all that money is buying something useful (especially as some state regulators continue to challenge smart meter benefits, like in Connecticut recently). And bingo: A treasure trove of data for creative app developers – who previously had to come up with their own screen-scraping code for each individual utility – in record time, only a few months after Chopra challenged utilities to adopt the scheme.

Green Button data is a good start, and I’m excited to see what the likes of Opower, Efficiency 2.0, Simple Energy, and others do with it. To achieve the potential, however, I think two more ingredients are needed:

  • Standardized rate plan info. Utility rate plans are byzantine structures that change frequently. Without knowing what rate plan a customer is on and what its specifics are (critical peak tariff? demand charge? prepaid?), apps can do little more than regurgitate back hard-to-parse kWh consumption data. Nearly every app focused on energy consumption has this need; I hear pitch after pitch from start-ups who have to address it by burning resources to reverse-engineer the same utility bills. So I’d like to see the U.S. CTO’s bully pulpit used once more, this time challenging utilities to develop a standard data schema for rate plans and to provide up-to-date descriptions online in machine-readable form.
  • Incentives that can compete for mindshare. A couple weeks prior to the Green Button launch, SDG&E announced the results of its pilot Biggest Energy Saver competition (patterned after a similar effort launched earlier in Texas). The good news is that consumers who interacted with a social game as part of the competition saved twice as much as those who didn’t: Score one for energy apps. The bad news is that the winner – one Erica Faunce of Lakeside, CA, who cut her electricity usage by a whopping 46.5% – won… a laptop. In my view, more heavyweight incentives will be required to make apps based on Green Button data engage consumers en masse – we’re competing with Powerball for attention here!

A week after the Green Button reveal, Chopra announced that he’s resigning from the U.S. CTO position (he’s tipped for a VA lieutenant governor run). Aneesh got a ton done in a short amount of time, aided in no small part by his outsized personality. Here’s hoping his successor is equally capable – and willing to take up the gauntlet for energy efficiency.

Posted in Consumers, Smart grid | Leave a comment

Q&A Roundup

I’ve gotten a ton of feedback on the “state of cleantech VC” series – thanks so much to everyone who reached out! I’ve found myself discussing some of the same points with different people more than once, so I figured it would be useful to follow up on them here.

Q: Do you have a PDF of this stuff? No time to read except on a plane.

Sure, why not – here’s the “state of cleantech VC” series in handy-dandy PDF format. I’m also adding it to the newly inaugurated “tools” page of this blog.

Q: You should really use acquisitions for the exit analysis instead of IPOs, because a lot more VC-backed companies get acquired than go public.

I’d love to do this, but I don’t have the data. I was trying to look at “successful outcomes as a percentage of companies invested in,” for both VC-backed companies overall and for cleantech specifically. While the National Venture Capital Association tallies up total VC-backed acquisitions, I don’t know anyone who does this just for cleantech. Further, only a subset of acquisitions count as “successful outcomes;” undesirable fire sales are common. Because acquirers tend to keep their purchase prices secret, we can’t tell the difference without prodigious research.

Q: The post on share price trajectories was interesting. Is there some kind of template out there that I could use to model the impact of raising money at different valuations?

As you can imagine, this one came primarily from first-time entrepreneurs. When I sit down with folks like this to explore a financing, we typically start with a generic Excel template that has all the formulas pre-wired. So go ahead, have at it with my cap table template (also hitting the “tools” page).

Q: Your third post seemed to indicate that raising money at a high valuation is bad for entrepreneurs. How on earth can that be?

Definitely not the point I was trying to get across. Raising money at a higher price than before is a great thing. The problem arises when you raise money at such an extraordinarily high price that any subsequent round is likely to be down – which selectively dilutes common shareholders (i.e. founders and employees). Get it? Higher valuation = great, stratospheric valuation = dangerous (unless, of course, you’re certain that it’s the last money you’ll ever raise). Which leads us to the last topic…

Q: It’s incorrect that down rounds selectively dilute common stockholders. As long as the price at the end is the same, it’s just a question of whether you want to take your dilution now or take it later. It’s the previous investors who get hurt in a down round, not the management or employees.

No! No! Wrong, wrong, wrong!

I heard this from a couple of first-time CEOs who have successfully raised venture money, and I was frankly surprised by it. The missing element is the impact of anti-dilution provisions which are standard terms in nearly every institutional financing.

In the event of a down round, anti-dilution provisions retroactively reset the share price of previous investment rounds to a lower value. This effectively issues new shares out of thin air to the preferred shareholders who invested in those rounds, but not to the founders and employees who own common stock. So while all existing shareholders get diluted once in a down round (by new money being raised at a lower price), only the common shareholders get whacked a second time (by the anti-dilution provisions).

I think entrepreneurs tend to overlook these very standard terms (until they bite!) for two reasons. First of all, the language is obtuse. Here’s an example from a recent term sheet that crossed my desk:

Antidilution Provisions: The conversion price of the Series A Preferred will be subject to a broad-based weighted average adjustment to reduce dilution in the event that the Company issues additional equity securities (other than shares reserved as employee shares described under “Employee Pool” below and other customary exclusions) at a purchase price less than the applicable conversion price.

The second reason is that these provisions are frequently implemented in the form of a “conversion price adjustment” which only manifests itself upon liquidity. In this scenario, no new shares actually get issued at the time of the down round; instead, when the company goes public or gets bought down the road, the anti-diluted preferred shares get converted into a greater number of common shares than originally specified. This delayed impact makes the effect easy to overlook.

Too little is written about this topic, but if you’d like to dig further, this Startup Company Lawyer post works out a case study, and this post by Brad Feld explores how anti-dilution provisions affect company control.

(A final point mentioned by a VC friend: If you’re a senior management team member and you preside over a big down round, your job security may face as much risk as your ownership percentage – perhaps the greatest reason to avoid these events.)

Posted in Numbers, Venture capital | Leave a comment

The State of Cleantech Venture Capital, Part 4: Parting Thoughts

(A version of this post also appeared at GigaOM.)

tl;dr: We’re in the early innings of a long ball game.

This week we’ve analyzed the state of cleantech venture capital and used data to discern myth from reality. In summary, we’ve found that:

  • There will be sufficient late-stage capital in the next few years to feed the ’06-’08 “baby boom” of cleantech start-ups, but there may be a corresponding dearth of Seed/Series A money.
  • Contrary to conventional wisdom, cleantech VC is not sucking wind compared with VC overall. The two are performing about the same on an interim basis and cleantech investment has actually overdelivered on IPOs.
  • Half of successful cleantech start-ups stumble on the way to the finish line, enduring a down round that disproportionately hurts founders and employees. Entrepreneurs should approach fundraising with a long-term orientation and be wary of sky-high valuations.

A few parting thoughts as I muse on these points:

  • It’s still really early. Cleantech venture capital investment only became substantial in 2006, and the average VC-backed start-up takes eight years from founding to exit. That means the real fruits of this labor are yet to come. If there is ever to be a cleantech IPO boom, one would expect it to start in the middle of this decade.
  • Limited partners should reassess early-stage cleantech. More and more cleantech venture capital is earmarked for late-stage growth equity deals. As a result, these investment rounds are likely to engender price competition that depresses returns. In contrast, Seed/Series A cleantech financing looks to be cyclically underserved, and the enhanced return profile that accompanies scarcer capital could help offset early-stage technology risk. If I were at an LP institution right now, I’d be looking for the sharpest early-stage cleantech investment team that can zig while most investors zag.
  • New company founders should weigh alternatives to VC. While CEOs at late-stage firms will have adequate financing options in the next few years, new cleantech founders will find themselves fiercely competing for capital. Think of it this way: Since 2009, about 50 cleantech ventures per year receive Seed/Series A funding. Do you want to place all your bets on being one of the fifty? There are plenty of other underexploited options for cleantech entrepreneurs – including grants from agencies like ARPA-E during initial technology development (case study: FastCAP Systems), funding from a large corporation in exchange for a preferred license or other IP rights (case study: Liquid Metal Battery Corporation with Total), and early sale to a corporation with an executive job in the bargain (case study: Zensi with Belkin).
  • The optimal investment vehicle remains to be figured out. I pointed out in the second post that VC firms have, so far, mostly restricted their funding to companies that fit in the venture “box” – i.e., $10-30 million invested over the life of a technology company, all in equity, for an outcome in five to 10 years. By definition, this excludes big-payoff categories with mondo capital requirements (like nuclear fusion), fields that have acceptable capital needs but stretch the timeframe (like advanced materials), adjacent investment opportunities in cleantech value chains (like land deals for biofuel feedstocks), and financing the deployments of technologies versus the technologies themselves (an odd one, since more value tends to get created downstream). All of these omissions leave money on the table. A number of VC firms – mine included – are blazing a trail by shaving these square pegs for the venture model’s round hole, and flexible alternative investors like family offices, superangels, and corporations have a window of opportunity exploit. Despite this, I can’t shake the suspicion that a truly purpose-built cleantech investment vehicle lies in the future, not the present.

Thus ends our whirlwind tour of cleantech venture capital. Studying a fuzzy topic like this keeps a person humble, because most of the predictions you make are likely to be wrong! Please don’t hesitate to contact me with your own, especially if they differ.

Posted in Numbers, The State of Cleantech VC 2011, Venture capital | 2 Comments

The State of Cleantech Venture Capital, Part 3: The Companies

(A version of this post also appeared at GigaOM.)

tl;dr: Half of successful VC-backed cleantech start-ups stumble along the way. Entrepreneurs who raise big financing rounds at sky-high valuations can end up shooting themselves in the foot.

Most cleantech start-ups require a lot of capital. The average VC-backed company that goes public in this sector raises about $120 million through five rounds of financing along the way.

Every time a CEO seeks new investment, she walks a tightrope. On one hand, she wants to raise capital at a high valuation to minimize dilution. On the other hand, she doesn’t want the valuation to be too high, because if the company loses steam it risks a “down round” ahead – i.e., raising the next round of capital a lower share price than before.

Down rounds dilute the ownership of common shareholders (the founders and employees) quite substantially unless they’re explicitly protected. This is because of anti-dilution provisions designed to protect the existing investors, who hold preferred shares; these terms are part of nearly every institutional financing and are too complex to discuss here. Down rounds can dilute those investors as well, but as long as they can pony up their share of the incoming cash, they can maintain their level of ownership. Common shareholders don’t have this luxury.

Conventional wisdom holds that CEOs should strive for the highest possible share price every time they raise money, because their number one job is to prevent dilution, and down rounds are rare events that only happen in mediocre companies.

A while back, I decided to test whether this actually holds true in cleantech. It’s particularly germane in this sector because of the large amounts of money that must be raised and the many rounds of financing required to reach the finish line.

To perform this analysis, we need time-series data on the share prices of privately-held, VC-backed cleantech start-ups. “But Matthew,” you protest, “private companies rarely publicize their valuations, and they certainly make sure to keep down rounds a secret!” In nearly every case, you’re right, but there’s one instance in which companies are legally required to publish a historical record of their private share prices: When they file to go public. In a happy coincidence, we can presume that companies angling for an IPO have also had some measure of business success. So by examining the SEC filings of public and aspiring-to-be-public cleantech start-ups, we can determine whether successful companies always raise money at ever-higher prices, or whether they tend to stumble along the way.

From trawls through SEC filings, I’ve been able to identify 24 such VC-backed cleantech companies for which private share price histories can be reconstructed – 15 that have gone public since 2000, and nine more that have filed an outstanding S-1. (A handful of additional companies didn’t make the cut because they went public on a foreign exchange with different reporting requirements, they raised only one round of private financing, or they had a really complicated history that I couldn’t untangle.) Here they are:

You can chart the share price trajectory of each of these companies on a line chart. The x-axis of the chart is time in months, and the y-axis is share price. Each inflection point on the chart represents a fundraising round (Series A, Series B, etc.) at which capital was raised. The lines between the inflection points represent the intervals of time between fundraising events. For the companies that have filed an S-1 but not yet gone public, the last point on the chart represents the share price at the most recent private financing. For the public companies, the last point is the share price at the IPO.

Looking across these 24 companies, three patterns emerge:

The first pattern is the build, where a company raises successive rounds of financing at steadily higher prices with no single round standing out. This is a good situation for the entrepreneur because it minimizes dilution. A venture investor with 20/20 hindsight would want to play this scenario by investing early and having sufficient follow-on capital to maintain the position, because the earliest investors tend to make the highest returns (both on a cash-on-cash basis and an IRR basis). Biofuels/chemicals company Gevo is a good example of a build:

The second pattern is the pop. In this case, a company raises capital at a modest uptick each time, except for one round where some extraordinary event sends the share price soaring. Entrepreneurs love this situation – particularly if the pop comes late, because it mitigates dilution right when big money is required! A venture investor, in hindsight, would want to get in just before the pop. Demand response pioneer EnerNOC is a great example: Its share price spiked 5.5x from its Series B-1 round to its Series C round, owing (as far as I can tell) to a really hot 2006 summer that broke demand response records.

The third pattern is the stumble. This is the position that no start-up CEO wants to be in, where something goes wrong – unexpected technology hurdle? cancelled deal? management shakeup? – and new financing gets raised at a lower price than before. Theoretically a venture investor with perfect judgment would recognize the dip as a buying opportunity, but in times like these it’s exceedingly difficult to discern a hiccup from free-fall. A123Systems is a good case study: The company attracted more than $100 million at a big uptick in May 2008 in anticipation of an IPO, but had to raise another $100 million at a lower price the next year when the IPO got delayed.

So, of these 24 companies, how often did each pattern occur?

  • Builds happened a third of the time.
  • Pops occurred one-sixth of the time.
  • Half of the companies stumbled. This is the most common pattern. And these are the successful companies that ultimately filed to go public; just imagine the others!

The conclusion for cleantech entrepreneurs: Watch what you wish for – you just might get it.

Every time you raise financing there will be a chorus of voices in the boardroom encouraging you to raise the biggest round possible at the highest valuation. Sometimes this will indeed be the right decision, and if you’re certain that it is, you should shoot for the stars. But often – according to this data, at least half the time – it won’t be. Should you get over your ski tips in valuation, you’ll set yourself and your team up for disproprtionate dilution in the future, where your only defense is the mercy of your board. These are the cases where it’s in your personal self-interest to raise the appropriate amount of money at a sustainable valuation from people that you trust – not the largest amount of money at the highest possible price from whoever’s willing to pay it.

Remember, the share price that really matters is the one at the very end.

Tune in tomorrow for some parting thoughts.

Posted in Numbers, The State of Cleantech VC 2011, Venture capital | 3 Comments