Grit · 18 May 2026 · From the week of 18 May
What It Takes to Build a Generational Company | Anduril’s Trae Stephens
These are notes on the conversation, checked against its transcript. The episode itself has the full discussion.
In brief
Trae Stephens, co-founder of Anduril and partner at Founders Fund, joins host Joubin Mirzadegan to discuss Anduril's early years, his path from Palantir into venture, and his views on founders, fundraising and team building. He describes Founders Fund as backing founders rather than sectors, and explains why he thinks very large early rounds hurt companies. Joubin argues that building a product does not make it sell itself and that momentum and velocity are a startup's oxygen, which Trae endorses, and Trae explains why complementary, spiky teams build the best companies. Much of the conversation covers faith, work and personal history, which is not included here.
For founders
- Trae Stephens prefers a normal early round over a mega round, because a very large seed or Series A makes the next round hard to set below that price.
- Joubin Mirzadegan treats momentum as the most important thing for a startup and describes a steady run of wins as more powerful than lumpy fundraising, a view Trae Stephens strongly endorsed.
- At Founders Fund, Trae Stephens says the key question at seed and Series A is whether the investor has conviction in the founder and their fit with the problem.
- Trae Stephens says the best companies are built from spiky, complementary specialists rather than well-rounded generalists.
- Anduril offered employees a tender at a higher price every 12 to 18 months, which Trae Stephens said lets them treat equity as compensation.
For revenue leaders
- Joubin Mirzadegan says a great product does not sell itself, and that companies which skip building sales machinery can see demand fade once early growth flattens; Trae Stephens agreed.
- At Palantir, deployment strategists went on site with customers in classified environments, because those customers could not describe their problems or data back to headquarters.
- Palantir ran a team called Leverage alongside deployment strategists as its sales function, focused on new agencies and expansion within existing customers.
- Joubin Mirzadegan says the belief that with AI customers will just come is like saying engineers no longer need to code because of Claude Code; selling doesn't stop, and people will still need to run things through to completion.
What was said 18, most useful first
Palantir set up a team called Leverage as its sales function, alongside the deployment strategists, and deliberately avoided calling it sales. Listen
Trae Stephens said the Leverage team was responsible for growing the business through new deployments with new agencies and through horizontal expansion inside existing customers. He said Palantir went out of its way not to use the word sales. He said that when he was there the company had no commissioned sales, which he believed had changed since.
“We stood up a team called leverage Which was effectively the sales team for Palantir”
Listen to the episode Sales team, hiring & comp Link to this Report a problem
Very large early rounds, such as a $60 million seed or a $100 million Series A, are bad for companies. Listen
Trae Stephens said raising a very large round means the next round cannot be smaller, and that a mega round is only justified if the price is so high that the company needs runway to catch up. He said phasing in capital more patiently lets the price be remarked over time so employees see appreciation, and keeps the price low enough to grant options at a lower strike price. He said this also serves the company, not only investors.
“I don't like the mega rounds especially at the early stage. Like I think it's bad for business.”
Listen to the episode Fundraising & investors Link to this Report a problem
Anduril presents employees with a tender offer at a higher price every 12 to 18 months, which Trae Stephens said lets them treat equity as compensation. Listen
Trae Stephens said Anduril was intentional about offering a tender at a higher price on that cadence. He said this lets employees see their equity as nearly as good as cash, and that it changes how they think about retention and recruiting peers. He said a company should not stall under the overhang of a very large valuation or round.
“we actually want to make sure like on a you know every 12 every 18 months that we're presenting a tender opportunity at a higher price to our to our team”
Listen to the episode Hiring & team building Link to this Report a problem
A company that has raised a very large round loses negotiating leverage with candidates and customers, because they know it has no cash constraint. Listen
Joubin Mirzadegan said that after raising a very large seed round, a company can throw stupid offers at people without needing to negotiate, but it has no leverage because candidates know there is no cash constraint. He said this compounds negatively with customers as well.
“All of your candidates know you have no constraint of cash, so you have no leverage to actually be able to negotiate.”
Listen to the episode Hiring & team building Link to this Report a problem
The best companies are built from spiky, complementary specialists rather than well-rounded generalists, and Anduril's leadership is organized this way. Listen
Trae Stephens described the X-Men framing Shyam Sankar used at Palantir, where people use unique superpowers together and fill each other's gaps. At Anduril he said Palmer is exceptional at idea generation, building first product iterations and communicating vision, but does not run quarterly finance reviews or gross margin decisions, and that the company would not be better if he tried. He said the executive team is deep enough that each leader can focus on a specific area.
“they're not built of a bunch of like really well-rounded jack-of-all-trades people.”
Listen to the episode Leadership & culture Link to this Report a problem
Building a great product does not make it sell itself, and companies that skip building sales machinery can lose demand once early growth flattens. Listen
Joubin Mirzadegan said technical founders often assume a great product sells itself, and Trae Stephens agreed this is not true. Joubin described a COVID-era pattern where growth looked strong until it turned out no one was buying without active selling. The conversation concluded that when the curve flattens and the machinery has not been built, the company is in real trouble.
“it's actually even worse if you build it and they do come, because eventually they'll stop coming.”
Listen to the episode Sales process & deals Link to this Report a problem
Palantir sent deployment strategists on site with cleared customers in classified environments, because those customers could not describe their problems or data to headquarters. Listen
Trae Stephens described his early role at Palantir as a deployment strategist, where engineers went in with the customer, sat with them, understood their problem, and fixed bugs and added features on site. He said the reason was that in classified environments, customers could not articulate their problem or data back to the home office in Palo Alto, so people with clearances had to go in and address the real problems at that site.
“they can't articulate back to home office in Palo Alto what it is their problem is or what their data looks like because it's classified.”
At Founders Fund, Trae Stephens says seed and Series A investing rests on conviction in the founder, not on a sector thesis. Listen
Trae Stephens said the firm invests in the person it believes is most likely to pull the thing off, and asks whether the investor believes in the person and their founder-market fit. He said Founders Fund has no list of industries it works through, and that the question is whether this is the right person to solve this particular problem. He said growth-stage investing involves more data, while at venture stage conviction in the founder is the main thing.
“at venture, when you're making seed series A sorts of bets, you just need to have conviction in the founder. That's it.”
Listen to the episode Fundraising & investors Link to this Report a problem
The reasons venture firms give founders for passing are often fabricated. Listen
Trae Stephens said that the honest reason, that an investor does not believe the founder will pull it off, is a hard conversation, so firms give out-of-the-box answers such as TAM, competition, or the economics of the check. He said the better test is whether the investor believes in the person and their founder-market fit, and if so, that is where the bets should be concentrated.
“Most of the arguments that venture funds will return to a founder when they're making, when they're passing are just completely fabricated.”
Listen to the episode Fundraising & investors Link to this Report a problem
Founders deserve investors' full attention, and he takes only a notepad into pitch meetings to stay present. Listen
Trae Stephens said he often has three or four pitch meetings a day and is unlikely to have looked into the companies in depth, so he reads a short overview before each meeting. He said founders on the other side have often spent weeks stressing over their deck, so the imbalance is large. He said he does not judge others who use laptops, but that being fully present has made his pitch meetings feel better.
“I've got like, 30 seconds of prep and they've got like 30 days of like thinking about this”
Listen to the episode Fundraising & investors Link to this Report a problem
Momentum is the most important thing for a startup, and that a steady run of wins is better than lumpy fundraising. Listen
Joubin Mirzadegan described stacking product wins, customer wins, fundraising wins, media wins and candidate wins, with each win helping the next. He said lumpy fundraising makes that harder, and that less money with a partner you are excited about, at a fair valuation, lets you show progress in six months and then raise again. He said this has to be repeated forever. Trae Stephens said he couldn't agree more.
“startups are like the oxygen of them is momentum and velocity”
Listen to the episode Strategy & market Link to this Report a problem
Anduril did not reprice its rounds to clear oversubscription, because Trae Stephens said that would have reduced velocity. Listen
Trae Stephens said every Anduril round was oversubscribed, and that the company could have adjusted the price to match supply and demand. He said that would have optimized for minimizing dilution and decreased velocity, which he does not think is the right call in most cases.
“Oh, yeah, yeah for sure like the Velocity was the name of the game and you know for us it was we were oversubscribed to every round”
Listen to the episode Fundraising & investors Link to this Report a problem
Anduril's path to a high valuation never felt inevitable, and that it felt hard the whole time. Listen
Trae Stephens recalled that in the first year Joe Lonsdale told him the company would be worth tens of billions of dollars, and he told Lonsdale to slow down because they were trying to do something incredibly hard. He said Palantir's offer letter had equity scenarios up to $50 billion that everyone found ridiculous at the time. He said only the most irrationally exuberant founders would believe in that kind of inevitability.
“It never felt inevitable. It felt hard the whole time.”
Listen to the episode Leadership & culture Link to this Report a problem
Roadrunner is in a small, unglamorous category where VCs he met cited TAM, and that he thinks the market is misunderstood. Listen
Joubin Mirzadegan said the CPQ category is boring, and that unless you come from sales you have probably never heard of it, while engineers building companies often do not know it either. He said he believed the problem was upmarket and required access to enterprise customers. He said investors he met with pointed to TAM, and that he has plenty of opinions on why the market is misunderstood.
“Like, there's a reason why nobody's done this. It's TAM constraint.”
Listen to the episode Strategy & market Link to this Report a problem
Trae Stephens questioned whether the number of companies worth investing in has grown in line with the explosion of venture funds. Listen
Trae Stephens said that 30 to 40 years ago there were only a few venture funds, and that today there are thousands. He said he does not think that math makes sense, asking whether anyone really believes there are a thousand times more companies worth investing in. He described a market imbalance in which too much capital is chasing too many startups.
“Do we actually believe that there are 1000x more companies worth investing in?”
Listen to the episode Strategy & market Link to this Report a problem
Trae Stephens predicted that enterprises will tighten their discipline on SaaS spending rather than spend exponentially more because more SaaS companies exist. Listen
Trae Stephens said the enterprises that spend meaningful dollars on SaaS will not multiply their spend because there are more vendors. He said they will tighten up and maintain discipline, and that he feels there is a market imbalance. He described competition as the killer of opportunity in Silicon Valley and said there is more of it now than ever.
“They're not going to spend exponentially more on SaaS because there's exponentially more enterprise SaaS companies”
Listen to the episode Strategy & market Link to this Report a problem
The belief that AI means customers will just come is like saying AI makes coding unnecessary, and that people will still need to run deals through to completion. Listen
Joubin Mirzadegan cited a tweet from someone on his team saying we wouldn't tell engineers they no longer need to code because Claude Code exists. He said the same applies to sales, which does not stop, and the conversation concluded that people are still needed to run things through to completion.
“It's the same thing with sales. Like, you know, like, this doesn't stop selling.”
Roadrunner plays a 'meet me' sound when users log in, a feature Joubin Mirzadegan insisted on, and Trae Stephens described audio cues as part of a product's aesthetic. Listen
Trae Stephens said Palantir's Akash Jain signed off conversations with 'meet me' and suggested Roadrunner do the same. Joubin Mirzadegan said it already plays on login, as the one feature he insisted on despite rarely asking for features. He said an early customer demoed it to all of sales with the volume up and was startled. Trae described Anduril designer Jen Bucci's point that audio cues, like the Netflix sound or the Slack tone, are part of a product's aesthetic.
“please just add the meet me when you, when, when somebody enters road runner”