Steve Blank of Adjunct Professor at Stanford joins Nick to discuss Maintaining U.S. Dominance, Navigating Defense Tech, Prime Obsolence, and Why Your Startup is Likely DOA. In this episode we cover:
- Changes in Product Development and MVPs
- Impact of AI on Startup Success and Founder Mindset
- Common Missteps and Digital Twins in Startups
- Disruption and Adoption in Enterprise Software
- Fundraising and Venture Capital in the AI Era
- Defense and National Security Innovation
- Challenges for Traditional Defense Contractors
- The Role of Dual-Use Startups
Guest Links:
The host of The Full Ratchet is Nick Moran of New Stack Ventures, a venture capital firm committed to investing in founders outside of the Bay Area.
We’re proud to partner with Ramp, the modern finance automation platform. Book a demo and get $150—no strings attached.
Want to keep up to date with The Full Ratchet? Follow us on social. You can learn more about New Stack Ventures by visiting our LinkedIn and Twitter.
Speaker 1 0:02 Nick, welcome to the podcast about venture capital, where investors and founders alike can learn how VCs make decisions and reach conviction. Your host is Nick Moran, and this is the Full Ratchet.
Nick Moran 0:17 Steve Blank joins us today from his hidden retreat on the coast of California. He’s an adjunct professor at Stanford and co-founder of the Gordian Knot Center for National Security Innovation. Steve is one of the key architects of the Lean Startup movement, and helped develop programs like the NSF Innovation Corps and Hacking for Defense. He’s also the author of The Four Steps to the Epiphany and The Startup Owner’s Manual. Steve, welcome back.
Steve Blank 0:46 Thanks for having me, Nick. Gonna be fun.
Nick Moran 0:48 It’s been a while — probably five years — and it’s been a while since the early days of the Lean Startup movement. So let’s just get right into it. Do those principles still hold up in the sort of AI-native tech environment we find ourselves in?
Steve Blank 1:04 Normally I would give you an engineering answer, which is a one-bit yes or no, but I’m going to give you a business school answer: it depends. And the answer is it depends, because I think it’s mostly over. Some of the core principles — you need to get out of the building and discover things, and I’ll explain what those things are — but boy, the world has changed. And if you haven’t been paying attention, you’ve probably been under a rock. Lean, which lasted 25 years in its current incarnation, is about to undergo radical transformation, along with the nature of how we’re building not only software, but hardware, life sciences, everything. The methodology — it was just a framework to figure out what you should be doing and when, and how to be efficient with your time and money — that’s also going to have to change.
If I can, I’ll just describe some of the things I’ve observed, which I’m sure a good number of your audience are observing or actually doing right now. Part of the reason for Lean — the part I contributed to — was customer discovery and customer development: get out of the building, discover whether your hypotheses were correct, use agile engineering to build minimum viable products, search for product-market fit, and pivot. If you discovered things that didn’t match your initial hypotheses, you were no longer going to slave yourself to a business plan — which I used to describe as the best example of creative writing you’ll ever do, even outside of an English department — and instead use the Business Model Canvas to track your hypotheses as you went along.
Now, if we look at what’s changed — what the users are doing and what the product is — this notion of what I call product-market fit, I’m not sure is accurate anymore. The example I use is my last company was an enterprise software company. In enterprise software you’re building tools and products and features for users in an enterprise to do some kind of business process. Just imagine it was a tool for someone in supply chain — they’d have a dashboard, a screen, maybe connections to different silos in the company, and their daily work was clicking on things, analyzing what they should do. Is inventory low? Should I order something? Let me click the button, send out a solicitation, look at the bids that came back. That was a human being, and you were building features for a product to see if it matched their needs.
I’ll contend today that the final product doesn’t involve a human being at all, or not very much. The final product is not going to be a screen. It’s going to be a set of agents. The user interface might be a command line that says something like “manage and order inventory and let me know when it’s done.” All of a sudden, it’s not the search for product-market fit — it’s the search for agent-outcome fit. That’s a big idea. And it’s not like these agents magically appear; you have to build them and understand them. So customer discovery is still there. You now need to understand the business process and what kind of guardrails you want to put behind the agents. You don’t want to end up ordering 10 trillion paper clips and have the world’s resources consumed making paper clips. You do need to do discovery, but the whole notion of the role of the human being and what product-market fit is kind of goes away.
The other thing that’s radically changed — and I see it in my classes at Stanford — is that the minimum viable product used to be both an artifact you could show customers to demonstrate your progress and something you could show to VCs as a signal of how real you were. Now the mantra is, your cat can develop an MVP in an hour. You can spin up almost anything, and if it’s taking you more than a day, you probably went to the movies and slept in, because the tools out there are just absolutely incredible. It doesn’t mean that’s your finished product, but the MVP is no longer a test of your team’s competence for investors, and it no longer limits how many variants you can put in front of customers.
Implicit in this — and sorry, I’m giving you the course lecture; is this okay to keep going? — is what it does to the tool stack, the number of people you need in engineering, and the quality and type of people you’re hiring. Engineering used to be a serial process with a defined set of skills — you went through this entire DevSecOps stack, and it required skilled people from top to bottom. That’s no longer true. You still need some skilled people with taste about AI safety, guardrails, what the code is going to do in corner cases, whether it’s really going to over-order inventory or accept any price on orders placed. You need people who understand how to write agents — or rather, how to prompt agents — and how to put the right guardrails in place and connect to all these functions, because a good app will have multiple agents running in parallel, coordinating with a framework. I mean, Claude Code and OpenAI’s releases just leveled us up to a place that was unimaginable two or three months ago. That’s a different skill set than someone you’d have hired six months ago.
The consequences we could talk about are just pretty insane. This didn’t exist a year ago — well, we saw it coming a year ago — but the releases of Claude Code, the continual iteration of large language models, don’t look like they’re slowing down. They’re getting harder to build, but they’re certainly going to continue at scale. What was your question?
Nick Moran 8:20 Well, just to play devil’s advocate a bit. Something we’ve talked about for years, and others are talking about more now, is that the rate-limiting factor for a lot of scaling tech companies has been developers, and that’s changing significantly. But I would argue the rate-limiting factor for successful zero-to-one or zero-to-five startups is mindset, true customer discovery, creativity, and vision — and walking that path. Do you think that’s changed, Steve?
Steve Blank 8:53 I think the definition of success has changed, and then we can talk about the type of entrepreneur. Way back in the Stone Age, when I was an entrepreneur, success meant an IPO — an initial public offering on a stock market — because that was the only time you or your investors would make money. That’s laughable now. We would all eat ramen, take incredibly low salaries, and worry about burn rate — how much money you’re spending each month. I’m not suggesting that’s gone away for every entrepreneur, but it’s certainly gone away for the top 10% who are in AI. That’s no longer the game.
What’s happened now — and again, I’m not talking about every entrepreneur in the world, but at least the frothy top — is that the ability to create a product that generates substantive revenue and the ability to make a pile of money are completely disconnected. You could raise $100 million, never ship a product, never generate more than a couple million dollars of revenue—
Nick Moran 10:05 —take a secondary, seed round—
Steve Blank 10:07 —and get insanely rich. I’m not suggesting that’s the path, but I’m suggesting it’s a path. That it’s no longer available is completely irrational and is the indication of a bubble. But this bubble has been going on for a while, because AI truly is disruptive and transformative. It’s like the internet bubble I lived through around the turn of the century. The only claim I’ll make is we actually had over $100 million in revenue in our third year — it wasn’t just froth — but the valuations were equally insane, just off by about three orders of magnitude from what we’re seeing now.
So to answer your question: yes, being a founder isn’t a job. It’s the world’s worst job and the world’s best calling. Many times I get students asking, “Professor Blank, I’m trying to figure out which job I should take.” Almost every year I’d get the same kind of question with the same kind of result: “Well, what are your choices?” “Oh, I got this great offer from McKinsey.” “Great company. What’s your other choice?” “Me and my friends are thinking of starting a company.” “You already decided.” “Well, no, I want to talk about—” “Listen, McKinsey is a job. It’s a great job. But a startup is a terrible job. If you’re not called to this — if you don’t wake up in the morning wanting to work all day and sleep under your desk, if you’re not excited about the product and passionate about the change it’s going to make, if you’re chasing the vision and not the money — that’s what a founder is.”
A founder, to me — and I’ve been saying this for a while — is closer to an artist than any other profession. Founders aren’t accountants. Accounting is a job. Being an artist isn’t a job. You can’t split a room in half and say half of you are going to be accountants and half are going to be artists. It comes from inside. It’s passion-driven. Artists see things that other people don’t — so do founders. Artists hear things that other people don’t — so do founders. Artists have a passion to create, and more importantly, a personality that accepts — maybe without realizing it — that most art fails. Most songs are terrible. Most plays never get produced. Most novels go in the trash. But they’re resilient. They get depressed, they drink a lot or whatever, but then they get off the floor and do it again and again. That’s the nature of a true founder.
And let me exhale and say: that’s changed now. Those founders still exist, but it is possible to just be completely in it for the money, because the money is there at an absurd scale. You can say, “Well Steve, that’s nice — glad you folks exist. But let me tell you how I’m going to ride this AI hype train, spin this thing up, raise $100 million, take $10 million off the top before the next round, and go home. And I’m 22.”
Nick Moran 13:38 So Steve, I loved your latest blog post. The title is “Your Startup Is Probably Dead on Arrival.” It’s provocative — it’s a good one. We just discussed how AI is rapidly commoditizing tech stacks and changing the dynamic of company building. What is the most common misstep you see founders making at inception?
Steve Blank 14:03 The “dead on arrival” part is this: whatever assumptions you had when you started your company, it’s time to step back, take a walk around the block, and see if those things are still true. Unless you’re constantly watching the tool stack and the funding environment, whatever you believed more than a year ago is simply wrong — not because you were dumb, but because the environment is changing so rapidly. You really need to look around. Look at your engineers — are these still the people you need?
For software, it would take about an hour of consideration. If you’ve missed Claude Code or the equivalent, you really should be rethinking your company. But for hardware people who say, “Steve, that works for software, but I’m building silicon or hardware” — it has everything to do with you. If you’re not building a digital twin, you’re missing the whole point about customer discovery. You shouldn’t be waiting a year to put something in a customer’s hands. You should be giving them digital twins — think of it as simulation on steroids that you can now share with a customer. They could be playing with that twin and giving you feedback constantly while you’re still building the product. They could say, “We really aren’t using these top three features, but numbers four, nine, and twelve are what we’d pay for.” And if you’re instrumenting that, you can see it in real time.
And it doesn’t have to be a digital twin of the entire product. It could be part of a manufacturing process, part of the key circuit, part of something else. You might think, “Okay, I can imagine that for hardware, but it’s not going to work in life sciences.” Well, AlphaFold won the Nobel Prize for essentially doing 75 years of work on protein folding. We figured out the configurations of maybe 50,000 or 75,000 proteins in 50 to 70 years. AlphaFold found 200 million, and most of them are correct or closer to correct than what we had from X-ray crystallography. And now we’re multiple generations past AlphaFold — doing complicated proteins, protein-membrane interactions, whole-genome predictions. So it’s affected every field, not just software or hardware.
What’s interesting is that the tech world — hardware and software — lives today in a completely parallel universe from life sciences. If you’re in life sciences, you say “yeah, Claude, blah blah blah, but let me tell you about my tool set.” If you’re in software, you know nothing about next-generation life science tools.
I think people need to take a long pause. If you started your company before about June 2025, take stock of what’s around you — do an inventory of technology stack, people needed, funding available. If you’re not in AI, the available pool of VC money has gotten smaller, not bigger. Why? I hear about all these billion-dollar deals — guess what, they’re not going into your area. That’s a giant sucking sound pulling all that money into AI. And rightly so, because VCs are essentially a financial asset class that goes where the returns are, and the returns are all going into AI bubble valuations. If you don’t have AI in your pitch, you’re probably not going to get a meeting, or getting one is going to be a lot harder. There are contrarian VCs, of course, but VCs are driven by where they’re going to get the most return. It’s just pretty simple math.
Nick Moran 18:44 As we look at the founders that are building, one of the concerns I have is related to the abundance and the speed — the amount one can do, and do so much faster—
Steve Blank 18:58 It creates a lot of slop, right?
Nick Moran 19:00 A lot of slop, right? Because more isn’t always better. Better is better. Brad Feld was on the show and shared that quote. So my question is: what should founders stop doing immediately if they want to survive and thrive in this AI shift?
Steve Blank 19:18 What founders need to stop doing immediately is thinking that whatever was in their heads prior to this year is still true. We’re back to customer discovery. It’s literally like you’ve just been transported to the 22nd century. These tools are kind of like a time warp that delivered 22nd-century tools right into the middle of our workday. So we need to get out and actually start testing — not only our own assumptions, which I talked about, but our potential customers’ assumptions as well.
Whether you’re selling to a consumer — what’s the platform? Probably some kind of phone or consumer device, and those haven’t changed radically — or in the enterprise, there’s the Alan Kay quote: the future is here, but it’s not evenly distributed. Some CIOs and organizations are right there, AI-native, willing to adopt at scale. Others are still running client-server somewhere in the basement. I mean, the fact that the Artemis astronauts were using Outlook — did you read that? They had to call tech support from a couple hundred thousand miles away from Earth because Outlook was spinning up two instances. If I were Apple or SpaceX, I would have said, “We’re sending a rocket with a Mac to catch you up.”
So you could be excited about the tools and what agents could do, and yet your potential customers are freaking out about security, guardrails, and AI safety and won’t even let it inside the building. That’s the other thing I would be thinking about: while you’re so excited about what you could do and your demos are great, adoption might be a lot harder than you think — or a lot easier, depending on who and where you’re selling.
Nick Moran 21:54 Talk more about that — the customer profile. We just went through this SaaS-pocalypse, right? $2 trillion in value has been erased. Some say it’s an overcorrection, some say it’s warranted. Talk more about what solutions will be built internally, how customers will make the build-versus-buy decision, and what types of products have true defensibility versus not in this new world.
Steve Blank 22:28 Way before most of your listeners were around, when the Mac first came out — we take for granted multiple typefaces and bitmap screens now — it was a shock to the system. It allowed everybody to say, “What do I need a graphic artist for? What do I need these professionals making slides and newsletters for? I can do all that myself.” And the answer was pretty true for basic stuff — you could now make your own newsletter. When PowerPoint came out, graphics houses went out of business and you were creating your own slides. They were the ugliest things you’d ever seen. Some of them looked like ransom notes — you could use a different font for every other word. We actually call that the “ransom note” literature. So quality went down for the first two or three years, until professionals just started using those tools and companies started providing templates. We went through a world where quality went down, but the number of people who could create things scaled exponentially.
Nick Moran 23:57 Classic disruption, right?
Steve Blank 23:59 Classic disruption. When we had bespoke weavers in England and then set up looms, the quality of cloth probably went down by 10x — it was no longer 300-count whatever — but millions of people had access to incredibly cheap cloth. Those weavers didn’t entirely disappear; a small fraction still exist today as craftspeople. But the masses now have access.
I think we’re going to see every company experimenting with these tools — it’s already happening. If it’s not a core part of operations, it’s hidden in the basement, where people are playing “hey, look what I can do.” And by the way, I’m sure one of these experiments will go worse than a cyber attack — it’ll erase customer accounts or place runaway orders because it was an agent, and then multiple agents started talking to each other. I’m sure we’re not going to accidentally shut down the entire power grid — but we are going to see this era of experimentation.
The good news is I think that will be another harbinger of a disruptive transformation. You’ll see every company asking — the SaaS-pocalypse was really just investors saying, “Wait a minute, if I can spin up a version of Salesforce, or at least the core features I need, within a week or two, is that good enough? And what does that do to every enterprise software company?”
There are moats — not every SaaS company is going out of business — but every one of them is asking the same question: are our moats in distribution? Are our moats in regulatory things — did we have to get licenses in every state for a banking app? Can you spin up regulatory approval in every state? That’s still another year or two. Or, can you do medical diagnostics for radiology? Great — do you have the corpus, the database of 5 million radiographs, to train the large language model?
So data — proprietary access to data — is still a moat, though the bar is rising because large language models are ingesting a ton of what used to be proprietary. Regulatory moats, distribution access — those still matter. But the moat of “I have access to the world’s best programmers” is no longer what it was. Whoever thought the first people to worry about their jobs would be software developers?
Nick Moran 27:09 I mean—
Steve Blank 27:11 It never occurred to me. I always thought, “Oh no, it’ll be the low-level clerks.” No — it’s actually the people who write code.
Nick Moran 27:19 I was in Menlo a few weeks ago, and someone was telling me that the top two graduates from Stanford CS got paid over a million by the biggest tech companies, while the rest of the class is struggling to find jobs.
Steve Blank 27:35 And by the way, they’re wrong — it wasn’t over a million. There was another digit involved, seriously.
Nick Moran 27:42 Okay—
Steve Blank 27:42 Yeah. I’ve had students tell me — when the Meta and OpenAI bidding war was happening — they were getting offers for more money than I made over a good part of my career. Unbelievable. But you’re right: if you’re a commodity — and let’s leave the top 10 universities out of this — if you’re just in CS at some other university, you’re going to have a very different career than you might have expected. It’s kind of like when doctors became commoditized. When doctors had to work in groups and practices, and those practices started getting bought up by private equity — you’re no longer this independent individual at the top of the profession. In the 1950s and ’60s, being a doctor meant you made more money than almost anyone else. Now you still go through the same arduous training, the knowledge required has probably grown 10 or 50x, but it’s become a commoditized business. Software development has just undergone that same transition.
The one that blew me away, Nick — I have some friends, one of them runs the CTO office in the UK’s Foreign Office. I saw him about three months ago in March, and we were talking in a nice theoretical vein about how they’re going to implement AI. I thought, “Well, it’s London — maybe by the turn of this decade.” I just got back there in June, and what they had already developed blew my mind. Literally in three months. And I’m thinking, this is London, not Silicon Valley. Shame on me for thinking people anywhere are less capable. We just have more access to capital here. The diffusion of this innovation was immediate and obvious to anyone in a technical cluster — or even just a room of four people. You probably figured it out in Mongolia that the world we knew was over.
And there’s another barrier worth mentioning: I went and said, “Do you understand how you’re going to get tens of thousands of people in an organization that’s been around since the 1600s to adopt disruptive technology?” Then we had to talk about saboteurs, the diffusion of innovation, how things get budgeted, what they’re going to turn off and what they’re going to keep. That had nothing to do with the tech — it had to do with deployment. Deployment, as anyone will tell you in every disruptive phase, is completely uneven.
When we built our last company, we were fighting against client-server. People were happy with it, and we were saying, “No, no, there’s this thing called the browser — you can access your data from everywhere.” That was not a technology problem. That was an adoption problem. Does that make sense?
Nick Moran 31:36 It does, 100%. So, the way things are evolving so rapidly — I wanted your input on fundraising. I sent around one of our Series B startups the other day. It 3x’d in ’24, 4x’d in ’25, doing significant revenue. And the response I got from a very famous name in the valley was, “We look for more explosive growth at this scale.”
Nick Moran 32:09 I had a chuckle. But the classic T2D3 — triple, twice, double three times — growth benchmark no longer seems to be the gold standard. Folks are looking for tremendous, rapid scale. You and I have discussed the founder in their reality distortion field. Is the venture industry in a distortion field right now, or is this what we should expect persistently going forward?
Steve Blank 32:39 We absolutely — listen, I actually did one of these in a much earlier bubble. We went from $3 to $12 to $120 million in three years, and that curve would have continued but the bubble crashed and the company got sold. We went public at $120 million, and back then an $8 billion market cap wasn’t considered a lot of money. But there was a revenue curve — that hockey stick of explosive growth — and that’s what investors and founders should be thinking about.
When I say get out of the building and look at what’s changed, what investors expect from you is that you’re on that curve. It’s not that other companies are dumb or wrong, but they’re no longer on the curve that makes sense for an investor at this scale.
I remind people that when you’re in an innovation cluster, it’s not just the bright entrepreneurs on one hand — it’s the people with capital on the other. Without both, you just have one hand clapping. There are still only 10 to 20 places in the world where you can raise $100 million. You have to understand the impedance match between their business model and yours. When you’re raising money, the minute you take money from someone, their business model becomes yours.
What that means today is you need to understand what their model is. Three years ago they might have said, “Hey, a $100 billion valuation is really good.” Now they’re saying, “Stand in line with the small business guys, because I’m investing in things valued at hundreds of billions — what’s SpaceX going to IPO at, a trillion, $2 trillion?” Think about it — we used to talk about unicorns on this show ten years ago, and now we’re talking about trillion-dollar private companies.
One of my friends is a VC who owns a piece of Anthropic. That firm is valued at around $380 billion. OpenAI at $800 billion or whatever the number is. We’re talking about valuations greater than the gross national product of some countries. If you own 10% of that as a VC, you’re looking at an $80 billion return in a fund. Now you’re asking me, as a founder, to tell you my thing is going to be worth $100 million? Tell me it’s going to be worth a billion, and you’re happy generating $10 million a year? That’s what I call the impedance mismatch.
So part of “getting out of the building” — looking at new circumstances — is understanding what VCs are looking for. It’s nothing personal. It’s about where they’re putting their money. If you’re not doing AI at that scale, your potential funding pool has gotten smaller, not larger. You need to find the VCs who say, “These valuations are insane, we’re not going to chase them, we’re going to go after different returns.” Those people exist, but you need to make sure you’re not looking in the wrong place at the standard names. If a VC is already invested in any of the large language model companies, they’re probably not investing in you. That’s probably a good heuristic.
Nick Moran 36:01 Product-market fit, but equity-market fit — it’s a different cast of characters for different types of businesses.
Steve Blank 36:06 And that’s always existed. I used to have students who’d say, “Well, I’m building this great small business — I’m going to drop $2 million a year into my bank account.” “Great. What’s the problem?” “I can’t get Sequoia or Kleiner to pay attention.” “Well, no wonder — you’ve built a great small business, but that’s not venture-fundable at the scale you’re talking to.”
Understanding who you are — I used to build a taxonomy: lifestyle businesses, small businesses, venture-scale businesses, intrapreneurs, NGOs, etc. I’d now probably add AI hyperscale businesses to that taxonomy. It’s just about matching your interest in returns with theirs.
And again, some of these businesses are ones where you might go back to first principles of being a founder: can I generate enough cash by myself? AI now allows us to do some of those things. You don’t even need the funding.
We’ve gotten so accustomed in conversations like this to talking about venture capital that we’ve forgotten the basic principles — maybe a loan from mom and dad. How quickly can you bootstrap this? How far can you get it? Can you get it to a point where somebody might actually say, “This is worth playing with?”
If we believe what we’ve said — that this is disruptive in terms of headcount and everything else — that article about two kids spinning up a billion-dollar-market-cap business with $400 million in revenue from a GLP drug company, while I initially thought it was BS, it does capture what’s now possible that wasn’t possible a year or two ago. We’re back to full circle: in the ’50s and ’60s, most startups were bootstrapped because there was no venture capital industry.
Nick Moran 38:18 So Steve, I want to switch gears. While I have you, I want to talk defense. We don’t have a ton of time and it’s a big topic, but can you help frame the historical context on defense and how the US achieved its dominance, and what’s required so we don’t lose it?
Steve Blank 38:43 Yeah. A couple of years ago I asked myself that same question — you know, we were all building consumer stuff: Apple, Intel, whatever, in Silicon Valley. But I kept hearing this background noise. I started my career in the military, and my first company was kind of a halfway house between a military career and a startup. Everything I have to say on this, I put in a presentation called The Secret History of Silicon Valley. Go to my website, SteveBlank.com — there’s a tab called “Secret History,” a video, and more links than you’ll ever want.
Basically, Silicon Valley was not started by entrepreneurs building consumer devices. It was started by a Stanford professor who ran electronic warfare and electronic intelligence in World War II for the US government. He set up a massive 800-person lab at Harvard called the Harvard Radio Research Lab, which had nothing to do with Harvard, radio, or research. When the war ended, he took his best engineers and managers back to Stanford and set up a microwave, physics, and electronic warfare department, because every university during the Cold War was working on weapon systems. Stanford became a center of excellence for the Department of Defense and our intelligence agencies.
Then the Vietnam War happened and all classified work got moved off campus, which by accident started a whole new wave of defense startups. Silicon Valley has always historically supported defense — so has MIT, Johns Hopkins, University of Michigan — there was always a defense component.
What happened about five or six years ago is a couple of brave VCs said, “I think there’s a business here.” At the same time, we at Stanford stood up the Gordian Knot Center for National Security Innovation, which essentially said that commercial technology was going to have a major impact on the Department of Defense — they just didn’t know it yet. I got the head of the Office of Naval Research, Admiral Selby, to see the same thing. The DoD needed to really tap into what was going on in AI, machine learning, autonomy, access to space, biotech — all these things we now take for granted as coming from startups.
We started a class called Hacking for Defense, which is now in 63 US universities, 17 in the UK, and has just spun up in Germany and Poland, where students work on national security problems. There’s now an enormous amount of VC money going into defense startups that never existed before — Shield Capital, DCVC—
Nick Moran 41:54 —$20 billion annually—
Steve Blank 41:55 More than that. The floodgates have opened. Why? Because disruption. Look at what’s happening in the war in Ukraine.
Nick Moran 42:05 What happens to the primes? Do you think—
Steve Blank 42:08 The primes — for your listeners, these are the top five or ten defense contractors: Boeing, Lockheed, what used to be called Raytheon, Northrop, Huntington Ingalls who builds aircraft carriers, General Dynamics who builds the Virginia-class attack submarines. They historically built these Battlestar Galactica-scale systems — giant things that required thousands of people. It’s exactly like giant pre-AI software engineering.
Meanwhile, adversaries were building non-exquisite systems: drones that cost hundreds or thousands of dollars. You don’t need to sink an aircraft carrier — all you need to do is make the flight deck inoperable by peppering it and destroying the catapults. Now you have the world’s worst floating hotel rather than a floating air base. Or you could have drones flying right into aircraft revetments, or into aircraft parked without shelters. The whole notion of asymmetric weapons — no one talked about it at scale. The war in Ukraine — people are now shooting a million drones a year at each other, maybe more. We’ve decided to scale up drones, autonomy, all of this.
To answer your question: historically, the primes — like large organizations — were not designed to build small, cheap things. Their whole overhead cost was designed to bid on and work with the world’s most complicated acquisition system. I remind my students that if you know anything about selling to the enterprise, you’ll put yourself out of business in about an hour trying to apply those same lessons to the Department of Defense. The people who control the budget are in a separate organization — in Congress. The people who write the requirements are over here. The people who actually buy are over here. The people who deploy are over here.
The other major shift that happened in this administration is they basically brought in people who took 80 years of acquisition paperwork processes and shot them in the head. It’s essentially as if private equity acquired the Department of Defense acquisition system.
Nick Moran 45:07 Good thing or no?
Steve Blank 45:08 Awesome thing — if you believe we ought to have an efficient Defense Department that can keep pace with technology. Cerberus’s Feinberg is now the number two at the Department of Defense. While you have the ideological warriors heading the department, the person actually doing the work internally and reorganizing is a private equity guy. They essentially did a private equity roll-up — took all these silos, put them together, and are getting rid of tons of paperwork. Over 80 years, the paperwork became the product. Now the message is: the product is the product, not the paperwork.
The problem for primes is they became masters of the paperwork, and that was almost an insurmountable barrier for a startup — who do I call, I don’t have 100 lobbyists—
Nick Moran 46:05 —bureaucracy—
Steve Blank 46:07 Yeah. So primes aren’t going to go away. I’m seeing some of them doing all of the following simultaneously: one, waiting out this administration hoping things go back to normal. Two, hiring more lobbyists on K Street to try to change some of these rules. Three, trying to acquire startups to get some of their DNA — either acqui-hires or straight acquisitions. Four — and I think the best one — changing their business model. Probably the best example, and I teach with him so I’m biased, is Lockheed, which has given a billion dollars to their head of corporate VC, Chris Moran, who’s probably the best corporate VC I know of in defense or anywhere. He’s also my co-instructor on Hacking for Defense. His portfolio is very thoughtful — it aligns with Lockheed’s mission and also with what national security needs.
So there’s a whole other parallel world of defense startups. And if you’re in autonomy or AI, that’s an opportunity.
Nick Moran 47:33 Does that impact your view on dual-use startups trying to sell both commercially and to the government?
Steve Blank 47:41 When this was a new market, people said, “You should probably do both commercial and defense — maybe earn your chops in commercial and then take it to defense.” You don’t hear that much anymore, because the defense money is here. Take that basket of money. Now, if you can deliver — say, 1,000 drones or autonomous surface or underwater vehicles — that’s the opportunity. The caveat is that’s also a bubble. You can’t have 500 drone companies — or rather, you can have 500, but 495 are going to go out of business in the next five to ten years, or get acquired. Even the DoD can’t acquire 500 of them; they may acquire five.
So the moat there will be speed of delivery and speed of deployment at scale. That’s what we teach our students: not only figuring out product-market fit, but in the defense context, “mission-solution fit” — do you understand the mission, do you have a solution, and can you build and deploy at scale?
To help my students with the “who do I call” question — the phone book of who to contact in what is now called the Department of the War — I realized there was no phone book. Nobody was going to write one, so I did. There’s now a 400-page phone book with a 40-page go-to-market strategy document, with diagrams of how to navigate the system. It’s free on my website. Go get it if you’re interested in this market.
Did I answer your defense question? It’s a great, enormous parallel universe that didn’t exist until recently. The whole blog post about “your startup is dead on arrival” actually came from a dinner I had with someone who was building autonomous software for firefighting — an existing market with lots of players. At dinner, I realized he knew nothing about the market that had emerged since he started his company. His software wasn’t useless — but there are people right now, in Ukraine, already using autonomous robots for supply chain logistics under fire. You’ll see robots trundling with ammunition, autonomous systems for medical evacuation. The worst thing that happens in combat is you send ten people to evacuate one and end up with ten casualties. What if you had robots doing that? We’re already seeing that used at scale, not just in trials — and that didn’t exist 18 months ago.
My analogy to this founder was: you need to take a breath, get your head out of what you were doing — not because what you built lacks value, but because you need to potentially pivot, or at least understand that there’s an adjacent market that doesn’t have existing incumbents in it.
Nick Moran 51:33 Steve, if we could feature anyone here on the show, who do you think we should interview and what would you like to hear them speak about?
Steve Blank 51:41 You know, the world is undergoing multiple disruptions. We’re in the middle of a war — the first land war in Europe in 80 years. Technology in both the commercial world and the military is changing overnight. I’d be thinking about someone with perspective, maybe people who have a much clearer view of what the future looks like than I do. I’m going to pass on naming someone specifically, because if I knew who it was, I’d be talking to them. But I’d want to go beyond “here’s what’s going on today” to “what does it all look like going forward?”
The politics feel like the Gilded Age in the United States — the 1880s to 1900s. If you’ve read US history, it was a pretty miserable time: ethnic cleansing, pogroms, you name it. The world we lived in from post-World War II to the 1990s or 2000s was actually a unique period of US history that had never existed before, and it looks like we’re reverting to some of these other eras — now with competitors who have nuclear weapons and technology disruption on top of it. A whole set of things coming at us simultaneously.
Now that I think about it: there’s an economist whose blog I read religiously every day because the links are incredibly interesting. Tyler Cowen, C-O-W-E-N.
Nick Moran 53:33 I haven’t had him on the program, but I know of him.
Steve Blank 53:36 I’d be interested in Tyler’s view. It’s one of the things I read every day. And by the way, that’s an underrated skill set for a founder — intense curiosity about a lot of things. You hear about agility, tenacity, and resilience, but you don’t hear much about intense curiosity. It’s also a hazard for founders because it can be a form of ADHD, but if you’re not curious about a lot of things across a lot of areas, you’ll never do the pattern recognition that says, “Wait a minute, this looks like that market — maybe this is occurring here too.”
Nick Moran 54:26 That’s right.
Steve Blank 54:27 And by the way, Lean and getting out of the building could sometimes be confused with ADHD — “Hey, we just discovered this from a customer, now let’s pivot!” Which is why I tell people: put a 72-hour hold on whatever you tell engineering, or else you’ll thrash the organization. You can’t just say, “We just discovered X, now do Y” until you have sufficient data. Getting out of the building without any constraints is actually a denial-of-service attack on your own engineering group. That’s actually a big idea.
Nick Moran 55:06 That was good, Steve. Do you have any habits or behaviors that are a secret weapon?
Steve Blank 55:13 I read a lot — and this goes back to curiosity. I’m just hardwired to learn about a whole set of things: defense, physics, life sciences. To me, that’s the most amazing thing about being alive today, in the world of the internet and large language models. You can learn almost anything. That’s my secret weapon.
And probably the only real skill I have is that I’m good at assimilating a lot of data and seeing patterns in it. That was really what Lean was: it wasn’t invented out of thin air. People were already doing this stuff. It wasn’t like no one had ever gone out and talked to customers. It was actually recognizing a pattern between successful founders and those that failed, and then my work and Eric Ries’s work and Osterwalder’s work was simply giving tools and language to the best practices of what we’d observed. To put a bounding box around the most important things. That’s what Lean was about — that’s all it was. And it’s ironic that it took me three years to wrap my head around something I can now explain in 30 seconds.
Nick Moran 56:36 There it is. Steve Blank, the first AI agent. Well, there you have it. He is Steve Blank. Check out SteveBlank.com — the Secret History is my favorite tab, but there are many. Steve, thank you so much for joining us.
Steve Blank 56:52 Thanks for having me, Nick. It was fun.
Nick Moran 56:58 All right, that’ll wrap up today’s interview. If you enjoyed this episode or a previous one, let the guest know about it — share your thoughts on social or shoot them an email. Let them know what particularly resonated with you. I can’t tell you how much I appreciate that some of the smartest folks in venture are willing to take the time and share their insights with us. If you feel the same, a compliment goes a long way. Okay, that’s a wrap for today. Until next time, remember to over-prepare, choose carefully, and invest confidently. Thanks so much for listening.
