Can you systematically build $10B+ ideas?
How founders of iconic companies use systematic research to find and incubate ideas (e.g. SpaceX, Palantir, Varda, Base, Kurion).
Hey everyone,
Last week, I wrote a post called Not Boring Labs, a concept to incubate vertical integrators within the Not Boring ecosystem.
I actually DM’d Packy and he was kind enough to share some thoughts on the idea, which was awesome! It’s good reminder that cold-emails are very powerful and people are open to helping you if you can find a win-win.
Last week’s post argued for why getting good at incubation is strategic for VCs like Not Boring. This week, I’ll analyze the research methods VCs & founders use to incubate iconic vertical integrators (SpaceX, Palantir, Tempus), specifically around ideation.
The case for incubation
Last week, I argued that the largest venture outcomes in the next decade will be techno-industrials (i.e. vertical integrators type startups that focus on atom-based industries). If you believe this to be true, then incubation is an opportunity for a VC (like Not Boring) to accelerate this future and capture more upside.
Financial upside: By owning a stake in the companies it incubates, a VC not only participate in their financial success but also strengthens its position as a hands-on builder of vertical startups.
Deeper insights: Incubation generates unique, first-hand insights into the process of building vertical integrators—insights that no other firm can replicate. These insights deepen a VC’s thought leadership, attract more founders and investors, and solidify its reputation as the definitive resource for techno-industrials.
Today we’re digging into how to actually identify ideas during incubation, drawing on case studies from a number of iconic vertical integrators.
Unintuitive relationship between ideas and domain knowledge
Step 1 of any incubation is to find an idea.
How does a normal founder find a good idea if they’re starting from 0?
Elon didn’t work in aerospace before SpaceX ($350B). Eric Lefkofsky didn’t work in cancer diagnostics before Tempus ($8B). Brad Keywell didn’t work in heavy industrial equipment before Uptake ($2B). The founders of Palantir did not work in government prior to founding the company.
I mention these examples because it seems like vertical integrators are often built by industry outsiders. In fact, Packy observed a similar phenomenon for his portfolio of burgeoning vertical integrator startups (Base, Fuse, Hadrian).
There’s something else about Vertical Integrator founders I’ve noticed that is counterintuitive: they often don’t come from within the industry they’re trying to reshape.
- Packy Mccormick (Not Boring)
The traditional startup advice is become an expert for 20-years so you can understand the ins and outs of an industry enough to spot the innovation. But these examples suggest an alternative theory — good incubation is a result of picking an industry that hasn’t seen very much innovation and following a rigorous research playbook to spot the right idea. Notably, it seems this is the strategy of Sutter Hill Ventures, a VC that incubated Snowflake ($61B) & Pure Storage ($21B).
So what’s going on here?
How were these founding teams able to spot ideas that insiders could not?
What methods did they use to build the domain knowledge needed to innovate in an unfamiliar industry and mine for ideas?
Can these methods be dissected, understood, replicated and taught to the next generation of founders?
If these magical methods exist and can be taught, will we see a rise in successful vertical integrator startups?
Systematic ideation
My current working theory is that the best founders and VCs have built repeatable methods to identify promising markets, find good ideas and execute on the opportunities. Further, they are so good, they can repeatably build $1B+ companies. If this theory is true, then every aspiring founder should aim to learn and apply the “magical arts” of incubation to their own repertoire.
First, let’s separate incubation into two steps ideation and execution.
Ideation : Given an industry, how do you build domain knowledge and figure out the right idea?
Execution: Given an idea, how do you execute on the idea (recruit founders, find customers, build product).
Today’s post dissects ideation for iconic companies and provides insight into how the best founders do this. Unfortunately, great founders don’t write textbooks on their ideation process so I did my best to compile a list of resources and interviews on how non-domain expert founders approached research and ideation and eventually built $1B+ companies. Disclaimer : I haven’t worked directly with these founders so this is all based on secondary sources.
Per my interpretation, ideation can be segmented into 4 archetypes. Among these archetypes expert-interviews seem conducive to incubation.
4 research archetypes explained
Visionary
Visionary founders experience a personal pain point, which inspires their company. For example, Airbnb was born out of a pain of not being able to make rent. This style is probably the least conducive for incubation because it requires more serendipity than systematic process.
Examples
Airbnb
Dropbox
Canva
DIY
DIY founding teams pick a vertical first and aggressively network with domain experts to build industry knowledge and identify promising ideas. These founders often begin with their own network of contacts. Through interviews and ideation, they find a promising product direction.
Flatiron Health founders (Zach and Nat) picked cancer, a domain where they had 0 prior experience. To iterate towards an idea, they built a pitch deck, presented it to their healthcare connections. With each iteration cycle, they refined their pitch deck to communicate ideas that would sell and asked for intros. Eventually, they landed on a promising idea after 20+ iterations. Similarly, AtoB (Stripe for truckers) was in the middle of a pivot in YC and decided to focus on a old industry (trucking). They networked and talked to as many truckers as possible before identifying a good product to build (fuel card). They had no fintech or trucking experience.
Examples
Flatiron Health (Nat’s retelling of how Flatiron found its idea)
AToB (YC partner Jared Friedman talking about AtoB ideation)
Watershed (Generalist piece on Watershed they found PMF)
Consulting first
Palantir is perhaps the only example of a consulting first approach that I could find. When Palantir was founded in 2003, most startups did not sell into government and the founders had intuited that government could use effective software. The problem is they didn’t understand the pain points of government well enough to build a product that could sell, nor did they have the necessary relationships.
To build product intuition, they sold consulting services. The counterintuitive bet is that consulting knowledge would overtime translate into a compelling product.
Examples
Palantir (Nabeel’s review of Palantir’s strategy)
Learn from experts
This is probably the most promising form of ideation and it’s responsible for a number of hard-tech vertical integrators (SpaceX, Kurion, Varda).
Jim Cantrell (cofounder of SpaceX) recounted that Elon Musk actually did not understand rockets well when he first began. In Cantrell’s own words,
This remarkable because it suggests that Elon Musk’s superpower is not an innate genius for rockets, but the ability to become the person with the most context for an industry (i.e. a super brain that absorbs all the viewpoints of many experts in the field). Elon is arguably the best entrepreneur of our generation so it’s naive to expect to replicate these results, but it begs a question.
Can founders build superhuman context for a niche domain using the methods the same methods Elon used for SpaceX? Why or why not?
Although Kurion exited at a much smaller scale, it seems that Lux Capital founders (Josh Wolfe and Peter Hebert) used a similar approach to tackle the nuclear waste industry. In their own words,
What these examples imply is that incubation is a skill of using expert interviews to accumulate differentiated insights. By talking to 100s of expert, you will naturally build a perspective on the industry that is unique and valuable. Based on these interviews, I could only infer that expert interviews are key, but questions remain.
How do these founders approach customer-discovery?
What questions did they ask?
What prototypes (if any) did they create?
How did these founders incentivize so experts to talk to them without a clear product in the early phases?
Is any of this teachable to normal founders or does it a high-level of existing career capital?
Examples
Why normal founders can’t do this
The most effective methods for ideation in overlooked industries (prime candidates for vertical integrators) involve systematically learning from experts. On the surface, it seems so easy (talk to lots of experts, identify problems, build a big company).
Here’s the problem: most founders can’t do this.
A billionaire like Elon Musk can leverage their wealth, network, and reputation to command interviews with hundreds of experts. VCs like Lux Capital or Greylock also have massive resources to source expert interviews.
For the ambitious aspiring founder (e.g. SpaceX or Tesla engineer) looking to build the next great vertical integrator—these options are out of reach. To launch a vertical integrator, they would need to:
Pick an idea space (e.g., energy, climate, infrastructure).
Build their own network of experts and talk to 100+ people to identify the right problem.
Launch an idea, find early adopters, and iterate on the product.
Achieve traction while juggling a demanding full-time job.
This process is overwhelmingly difficult for anyone who isn’t already an exceptional outlier. Yes, some founders will brute-force their way through the DIY approach, but they are the exception, not the rule. The difficulty of doing research is the primary reason I believe so few vertical integrators are built. This difficulty also why I believe the majority of YC companies focus on B2B software problems (sales, CX, product) as opposed to building vertical integrators. My hunch is that founders pick easier domains like B2B SaaS not because they don’t want to build harder vertical integrators, but because the research required for vertical integrator has a high barrier to entry. If research were easier, I suspect many more vertical integrators would be built.
What’s next
The mission of this substack is to answer the question
How do we help normal founders anticipate the future, act on this knowledge and build meaningful, impactful companies?
We’ve uncovered a piece of the puzzle : great founders employ non-obvious research strategies to learn domains and identify promising startup ideas. So how do we push on this further? I believe the answer is to dramatically lower the barrier for researching new domains so that we can build more vertical integrators.
To illustrate this, take a look at YC’s most recent Request for Startups. It’s interesting to see that YC actually wants founders to go after very difficult domains (defense, space, cancer). But YC and other VCs are missing out on a key opportunity.
Unfortunately, the RFS isn’t particularly helpful.
YC’s RFS targets founders who have the preexisting domain knowledge in hard fields and the RFS is an ad to apply with the idea. However, YC (and other VCs) are neglecting founders who are interested in these areas but do not have the required domain knowledge. To serve this group of founders, YC and VCs would need to build products that lower the immense barriers for 0 to 1 product research in difficult domains.
Critics will say this is the job of the founder to figure out how to crack the research problem. I disagree. I believe there’s a way to take the lessons from the last decade’s incubations (Snowflake, Kurion, SpaceX, Tempus, Uptake) and package that into a product that trains founders to build their own vertical integrators. It won’t lead to a 100% success rate, but it will catalyze more vertical integrators and there is value to be captured there.
Admittedly, a product to make research easier is abstract right now. As I embark on a journey to build my next ambitious startup and I plan to figure out my own tooling for dong this research and convert that into resources that make it easier for founders to chase harder startups.
Thanks for reading,
Andy
p.s. If you found this interesting, please consider subscribing or liking.
p.p.s If you have ideas for researching and incubating vertical integrators, my DMs are open.
p.s. No LLM was used in creating this essay
Really interesting article!
This article really spoke to me. I’m very interested in starting a company targeted at US healthcare inefficiencies, but with exactly 0 domain expertise and network, what’s ahead of me is pretty daunting!