Why Deep Tech Startups Should Approach Services Revenue With Caution
Services Revenue Often Comes With Significant Opportunity Cost You May Not Be Accounting For
TL;DR
Deep Tech founders are on a completely different rail than their software and hard tech counterparts. The ambitious quests deep tech founders are funded to undertake come with many years of development and tens, if not hundreds of millions, of dollars in R&D before they can bring a minimum marketable product to market. But the prospective returns can be so significant that they knock history off it’s axis: look no further than SpaceX, OpenAI, Intel and Genentech (the first company real venture backed deep tech play which brought synthetic insulin to market in the early 80s) for four examples.
Despite the understanding that deep tech investments are pure plays, many VC firms that mentor and fund them want to hold them to the same metrics they hold SaaS companies to - like the traditional “5 years to 100M ARR hockey stick” revenue profile gate all too familiar to nearly a generation of YC graduates. This is likely driven by the fact that they are trying to shoehorn them into their own fund construction & return models they sold their own LPs on. This can lead to a push for early revenue from services unrelated to your core product. While this is wonderful in bootstrapped companies that are trying to stay alive and sometimes take a lifetime to build, in VC backed start ups where speed and focus is their superpower, this may send many early stage founders on wild goose chases for subcontracts, unrelated SBIRs and unrelated products that have nothing to do with the original play they sold to the investors.
Your company needs to be singularly focused on maturing your technology, incorporating it into a product and getting it to market, not on pushing for some unrelated services revenue to goose your financial metrics.
As a deep tech founder who has also fallen prey to this before and who chats with at many other deep tech founders on a daily basis and has had companies I invest in fall down this rathole too, I have seen lots of flavors of this same mistake. I thought it would be useful to talk a little about this common pitfall and why you should exercise extreme caution when pursuing revenue unrelated to the core problem and product you are trying to solve. Your company has to be singularly focused on maturing your technology, incorporating it into a product and getting it to market, not on pushing for some unrelated services revenue to goose your financial metrics. Product revenue is king and the sooner you get there, the sooner your company will be worth something and not sold for scrap. In other words, embrace the Pure Play.
Reiterating Perkin’s Law
For those who haven’t read my previous discussions on Tech Risk vs Market Risk I’ll start by reiterating Perkin’s Law in this context. Originally devised by Thomas Perkins of Kleiner Perkins fame in the 70s it is a core tenet of venture investing that “Technical Risk is Inversely Proportional to Market Risk.”
Deep Tech plays are a hope that a Science Fiction like technology brought to market can effectively create a long term monopoly for the company by creating something completely new and technically difficult to replicate. In his seminal work from 2014 Zero-To-One, Peter Thiel articulates this more broadly as the idea that “To build a great company that escapes competition, focus on creating the singularly new.” There is nothing more singularly new than something that hasn’t been invented yet!
Defining our terms GOOD services revenue vs BAD services revenue
I want to start the deep dive by making it clear that there is such a thing as good services revenue for deep tech companies: if it’s related to a pilot or sale of your product, or pre-work for an evaluation of your product, it’s good services revenue. Two real world examples of this include an SBIR in your product’s focus area (Epirus’s first SBIR win was a UAS pitch day with the Air Force where we proposed a drone mounted EMP solution), an evaluation contract to determine the benefit of your product for a customer (Spartan did a small evaluation contract with Uhnder for our software as part of our partnership early on) or a services agreement to support test & evaluation of your product or adapt it to someone else’s technology stack. These contracts are especially attractive if they come with selling actual A-sample units for evaluation. It’s entirely appropriate to gauge customer’s seriousness in buying your product by their willingness to put money on the table and if you are truly state of the art, there will be some services attached to it to adapt and integrate it.
Bad services revenue is when you take on engineering work unrelated to your core product. Examples of this may include taking on subcontract work where other companies are interested in leveraging unique expertise from your team for their own IRAD or CRAD contracts; Chasing SBIRs that have nothing to do with your core product or services work on product offerings from other customers unrelated to your roadmap. Another corollary to this is that is unpaid is work with other start ups where they are offering you equity in exchange for integration opportunities or signing up for partnerships where they have no ability to pay, but that’s for another time.
Deep Tech Takes 10x Longer to Go To Market - Smart VCs Know That
To really drive home the point on just how different deep tech companies (including Software deep tech companies like OpenAI) are versus traditional software companies that VCs are used to funding, I’ve put together a couple of tables. First let’s start with the Deep Tech companies who already have products in market. I’ve gone with both old and new here to make the point that this is era independent.
The average time to market for these deep tech companies is almost 69 months- almost 6 years on average to bring a first product to market! Contrast that with the 10 largest Software companies in the world shown in Table 2, which had an average time to market of a little under 7 months. So basically deep tech takes 10x longer to go to market than Software - no wonder investors want to see early revenue!
Average Time to Market for 10 100B+ Software Co’s First Product: 7 months
Average Time to Market for 13 selected Deep Tech Co’s: 69 months
With such long timelines the pressure is often extremely high for deep tech founders to show revenue- any revenue - early in their lifespan to unlock the next tranche of funding and prove to their investors that they aren’t just paying for a giant science project (even though by investing in a Deep Tech start up, you really are funding a giant science experiment and it’s probably a very Boolean bet). Enter the Services Trap!
The Services Trap
With product revenue a long way off, but (hopefully) possessing a team with some of the most talented and specialized engineers in your field, a lot of outside companies and government program offices will *love* to rent your big brains for some special IRAD project they either lack in house expertise to execute or can’t staff. The relationship often starts with you briefing them with your best and brightest engineers talking about your core product - this then leads into “that’s really interesting and super cool, but we could really use your help with this…”. This then leads to them telling you about some project they don’t quite have the expertise to finish or which perhaps some subset of your products envisioned features could solve for. They may even blow smoke up your hindquarters about the full market potential of their pet project and try to convince you that it’s worth your buy-in. Don’t fall for it. The opportunity cost is extremely high.
The reason for this is simple: in order for you to make enough money off these contracts to move the needle, you’ll need to put a significant part of your best resources on this and delay your core product development. To compensate, you will likely hire more engineers which subsequently increases your burn, making you hire faster then you planned, and now you have to find something else to do with them when the project concludes in a few months because your burn is too high. The cognitive load of managing this side project also comes at significant opportunity cost to your core product at a time when your company is still young and figuring out how to manage itself. So while on paper it may look like you’re making money, the drop in global productivity on what your investors funded you to do is harder to pin down but very real. In the long run, it will make your core product cost more and take longer to develop.
Big companies and national labs are more than happy to subsidize their under-resourced R&D budgets with your VC money. Even if you they have to pay a little bit extra for you to make a profit, you are likely much more efficient/productive than they are and come with less downside risk to their core capabilities if they need to turn the project off later (it’s easy to terminate contractors- layoff or redeploy internal employees you hired for short term projects, not so much). While it’s nice to get some money in the door - be wary of efforts that will distract your core talent from the mission your investors are paying you to succeed at: maturing your tech, finishing your first product and getting it to market!
The “Deep Tech as a Service” Gimmick
Another flavor of the above I’ve seen happen now a few times is actually self-inflicted. Recognizing that early customers are often willing to buy milk but not the whole cow and having heard from their MBA friends that “software-as-a-service” companies run high and very attractive margins in the form of subscription fees, many deep tech founders fall into the trap of thinking that they can build a lesser version of their product sooner and sell it “as-a-service” to perform some less spectacular task for early revenue. For an enterprising deep tech founder, realizing they may be several years away from selling their core product, this seems like a fairly attractive way to get money in the door. This is the “Deep Tech product-as-a-service” trap and it’s important to realize it’s a mirage.
There is a couple of reasons why these are mirages. The first is that they usually come with significantly smaller TAMs/SAMS/SOMs than the original deep tech play or are in market niches that represent contracting customer bases. Even if you win in this category, it will be a small fraction of the revenue and profits you would achieve by getting to your end product goal even months sooner by avoiding the distraction.
The second, your prospective customers probably have some existing baseline capability already in-house or that they are working with that you are merely taking overflow from - so the revenue will be chunky, intermittent and unreliable. Sure, your equipment may be better but they already paid for their own capex and they need to justify maintaining it - you will always be second fiddle until the internal capability is phased out.
The time to look into options like “as-a-service” is after you already have the “deep tech” part figured out and are looking to expand your beachhead, not to attack it on multiple fronts out the landing craft
The third and perhaps most important reason is that managing these side projects may cause you to hire a bunch of sales and operations people unrelated to your core product and build up a bunch of contract and overhead tackling to manage it. Most customers that want to hire you for this come with a lot of bureaucracy you don’t really want/need at your stage: things like security clearances you have to carry, more robust physical security (to manage their stuff) and cybersecurity requirements (to manage their data) and other hidden compliance costs to win this business. While many of these may seem sexy on the surface (“wow, they are clearing us into their program, they must really like us”) the experienced will see this as slowing you down and a distraction.
My advice is to avoid falling into this trap and merely take these as proof points that you are on the right track. The time to look into options like “as-a-service” is after you already have the “deep tech” part figured out and are looking to expand your beach head, not to attack it on multiple fronts out the landing craft.
Oops, I became an SBIR mill
Unfortunately I’ve made this mistake myself before or else we wouldn’t be talking about it here. Some founders in government oriented spaces get the bright idea in their head that rather than pounding the pavement and looking for government customers that could use their products specifically, then enticing them to work with you to buy or adapt your products to serve their needs through tailored SBIRs and OTAs, they should just apply for every SBIR where they think they may have a shot. Seems easy enough right? We’ll just throw everything at the wall and see what sticks!
Some of them even take pride in this, proudly talking about the unrelated SBIRs they’ve won. Some may even have dozens or even hundreds of these contracts: we call these SBIR mills. It’s a very typical offramp for deep tech start ups that fail to launch- indeed, some of the most famous SBIR mills began as tech start ups that failed to get product market fit - but were awesome at writing SBIR proposals!
Unfortunately the small size of Phase 1 SBIRs (some are as little as $50k with a maximum size of $150-250k in most cases) means that when accounting profit margins which are hard to calculate for a small and rapidly growing company, you are perhaps even losing money on net when you account for increased management overhead and SBIR production costs. Also your company is still at the top of it’s learning curve with respect to proposal production (if you want to move down that curve faster, may I suggest on of my angel investments: Procurement Sciences?) and writing SBIR proposals is cumbersome and time consuming. It’s therefore important to put SBIRs in the right light and be wary of applying for them in unrelated topics.
Another reason why this is a bad idea is for SBIRs bid properly vs product gross margins. If you are bidding competitive SBIRs properly, you should be bidding them at somewhere between cost to at most 20% gross margins (if they were cost plus you really couldn’t even go above 15% perhaps). Indeed, since true research is hard to pin down in terms of actual hours and materials spent, you run the risk of losing money. If you are losing SBIRs on price, it’s probably because you aren’t doing this. These are R&D contracts meant to help you win future revenue, not make your company profitable.
Putting your best engineering talent on bidding and executing unrelated SBIRs dilutes your team’s focus on your core product
Contrast this with product revenue where you should be aiming for 60%+ gross margins in the long run. So putting your best engineers on SBIRs you are bidding on - something the SBIR evaluation criteria, with provisions in it like giving you extra credit for having a principal investigator who is a PhD incentivizes - is diluting your talent focus from bringing your product to market. Steer clear!
Furthermore, smart VCs can see right through this and know that these unrelated SBIRs are a waste of time. They know that 99% of SBIRs do not transition to a program of record and wonder why you paying the opportunity cost to chase stuff not in your product’s roll out plan. They will adjust your forward twelve month revenue accordingly and try to focus purely on revenue from products. So all that trouble to boost your financial metrics with them will be all for not.
SBIRs aren’t meant to be a business - they are non-dilutive seed capital from the government to mature the state of the art. Don’t fall for the trap of thinking every SBIR you could win is a good idea.
The “My SBIRs Are Spawning 1000 Different Products” Trap
Deep tech by definition has a lot of very research oriented engineers as founders and early stage employees. These people are usually out of the box thinkers by nature: it’s one of the things that makes them great and what we love about them. However, much as how harnessed fire can fuel an engine but unharnessed fire can burn down a forest, these superpowers have to be harnessed and focused to do something great. Perhaps one of the most powerful things about Elon’s algorithm is that he ruthlessly focuses on simplifying the requirements and removing components not on adding new products to his business plan and complexity.
Once you go down the SBIR rathole of chasing unrelated contracts and - God help you - actually winning them, you are playing a very dangerous game. Your engineers will love being in a constant research loop because it’s new and interesting…and progress on your core product will stall out as you put out fires on and juggle a whole bunch of these new, poorly paying, contracts. Your engineers will even start to say things like “part of what makes DeepTechCo great is that we embrace new ideas and run with them and we’re creating tons of new innovative products.”
Around this same time, progress starts to slow on the product you were funded to actually create and the rest of your company starts to get impatient. The business development team - worried they aren’t going to meet their numbers- starts running off to market these “new products” which are mostly vaporware because, at this point, the core product appears to be stalling out and they don’t want to mark it zero on their bonus. Meanwhile your board starts to look like the guy in Figure 5.
SBIR mills from a VC standpoint are basically worthless: they can’t be sold to larger companies because all their profits are derived from contracts they have to cap their head count (at 500) to collect. Getting acquired by a larger co shuts off the SBIR gravy train - and SBIR mills have few if any transitionable products to scale. It’s deadweight in a VC portfolio. So essentially you become an acquihire, which is often a near write off.
Feeling your VCs pressuring you to getting back to spending their money on the original thing you sold them, you start to rein it in. This causes major pushback from the engineers from taking their toys away from them - before it was a fun place to work, now you say its time to focus and actually produce something! Wait, we have an actual boss again, what is this?
So now you’ve got morale problems on your hands, from killing non-core product efforts after they get started - and even worse after external customers are paying for them - is always harder than you think. The best solution: don’t spin them up in the first place! Focus on what you raised the money to do!
Treat SBIRs and Research Grants as non-dilutive revenue - not as P&L
One thing I’ve seen a few software companies I have invested in do that I quite like is that they will specifically call out their revenue due to services along side their ARR. For instance, I recently got a monthly update from one that said “we have $4.3M in ARR (4.6% is from services).” It’s ok for you to have services revenue and it certainly counts for something, but you shouldn’t think of it in the same way as product revenue. At the same time it can help extend runway. Honest deep tech companies should consider doing this too.
Wrapping it up: if you’re going to be a pure play, be a pure play
Philosophically speaking, Steve Jobs had a lot of wisdom on this (and so many other topics): “People think focus means saying yes to the thing you've got to focus on. But that's not what it means at all…Innovation is saying no to 1,000 things.” It’s important to be brutally honest with yourself about whether money in the door is related to your core product or not.
Don’t fool yourself into thinking that services revenue is something you need to be chasing at the expense of your core product. Exercise extreme caution if it’s off course from the core earth changing product you were funded to create. You’re a deep tech play because it’s hard and you need to try and get there as fast as possible, not get sidetracked trying to chase low margin and often illusory revenue from services.
As Napoleon put it “If you are going to take Vienna, take Vienna” - Deep Tech plays are ambitious and their investors should appreciate this. Don’t fall into the trap of looking for an off-ramp to being cash flow positive via services revenue before you get across the line with your end goal product
As Napoleon put it “If you are going to take Vienna, take Vienna” - Deep Tech plays are ambitious and their investors should appreciate this. Don’t fall into the trap of looking for an off-ramp to being cash flow positive via services revenue before you get across the line with your end goal product. Undisciplined services contracts can create distractions that drain your core team’s progress, burn cash even more quickly and can delay go-to-market even further. The harder, but more courageous, choice is to stay the course, tighten your belt on burn and other matters unrelated to your core products, and try to get the core product done sooner - that’s how you take off.