Navan's Ariel Cohen: What's Not Translating From $6.2BN to $2.8BN

Summary

• Navan went public last month at a $6.2B market cap and now has a $2.2B market cap. Ariel Cohen’s defense of the timing: the post-COVID capital structure was one driver, but a payments business “may end up not where you want to be” if it stays private too long, and enterprise customers “want to know that you’ll be around.” He started non-deal roadshows in April the same day the tariffs were announced — his conclusion is that timing a market is really, really hard; you follow through and run a good business.
• The market’s two mispricings, per Cohen: Navan looks like SaaS but is a consumption business — go-to-market cost and commissions hit the P&L today while the payoff arrives “in the years to come,” so investors see spend without the return (he claims “extremely efficient” GTM and churn without sharing numbers). Second, the blanket “everything software is dead” mood leaves “not a lot of patience to see the differences between the companies.”
• The current Navan AI-platform push began in a panic: roughly four years ago Cohen saw Sam Altman demo ChatGPT pre-release at a Napa conference, called his co-founder Ilan and said “If we will not build our own platform right now, we are so dead.” That platform (Cognition + the Ava agent) now handles the hardest support case in travel — during the JFK/NY airport shutdown weekend, Ava ran 55% of chats and worst-case call wait was 16 minutes.
• On build-vs-buy for customer-service AI, Cohen is categorical: neither Sierra-tier vendors nor raw Anthropic/OpenAI APIs are “even remotely close” to what a complex regulated vertical needs, because zero hallucination tolerance is the spec — send a traveler to the wrong gate and “you’re gonna also get a lawsuit.” The read-through for vertical AI investors: the cute LLM booking demos are “not even remotely close to being there.”
• Harry’s Salesforce pushback is the best exchange: distribution is all that matters, so an AI sales rep plus Salesforce’s reach makes it “a screaming buy.” Cohen’s counter — “I’ve never met a salesperson that told me that they like to use Salesforce” — and if users hate you, distribution only delays disruption, as he saw selling HP software (“Walmart to sell servers”). His proof that user love is the moat: in Navan’s history it lost six enterprise customers, and five of them came back.
• Cohen calls LLM infrastructure “commoditized” and “not an important discussion” — Harry counters that Harvey and Lagora both flipped from diehard OpenAI to diehard Anthropic on Opus 4.5 quality, the opposite of commoditization. Cohen’s honest non-answer: “I actually don’t have a clue” how an investor picks — Navan’s Cognition routes between models and many Edge transactions are with Google, while Anthropic’s dev-infrastructure lead looks “similar to what Microsoft did in the ’90s.”
• The vibe-coding data point worth trading on: co-founder Ilan vibe-coded Navan’s expense product in six hours over a weekend, and ERP integrations to NetSuite “are not [hard]. They are not.” Cohen’s implied map: app-layer software and integrations are now easier to vibe-code, while travel’s licensed, fragmented plumbing is “almost impossible to disrupt” — which redraws where moats (and engineering budgets) live.
• On Jason Lemkin’s line that no public-company CEO is happy: Cohen concedes share price “one hundred percent” correlates with employee morale and that staff refresh the stock app all day; his job is pointing them to the two-, three-, ten-year frame, Bezos-letter style. He checks the price twice a day, can’t sell anyway, and notes the same resilience test as COVID — when employees left for Zoom at its peak, he asked, “do you think more people than the people that live on this planet will use Zoom?”

Deep dive

1. Going public into the storm: why Navan didn’t wait

• Harry’s blunt opener — in a world where Stripe and Databricks push going public out as much as possible, “were you forced to because of the debt component?” Cohen concedes capital structure post-COVID was one reason, but stacks others: in payments, “there is huge, huge advantage to be a public company” in how you raise capital for that side of the business — stay private too long in payments and “you may end up not where you want to be.” Enterprise customers also value it: “They want the transparency… They want to know that you’ll be around.”
• On timing: Navan started non-deal roadshows in April on the same day the tariffs were announced and the market corrected. His takeaway — trying to time a window is “really, really hard,” so “we kind of decided to go for it… eventually we are running a really good business, so everything will get sorted out.”
• Asked the Bill Gurley price-to-perfection question, Cohen refuses the frame entirely: the IPO day matters less than whether Navan takes a market he defines as every frequent traveler — managed (Concur, Amex) plus the bigger non-managed side — with the real view starting “two, three years from now” once lock-ups, VC distributions, and “beat and raise… mechanics” wash out.

2. What the public market isn’t pricing

• Cohen’s diagnosis of the $6.2B→$2.2B ride: Navan has no clean comparable. It looks like SaaS but the model is consumption — “we only make money when you use us” — and all go-to-market cost lands today while the money comes in the years to come. “What investors see in our P&L is a huge investment in go-to-market, and they don’t see the immediate return.” He claims “an extremely efficient go-to-market and an extremely efficient churn profile,” while explicitly declining to share the numbers.
• The second mispricing is macro mood: “everything software is dead” gets applied indiscriminately, and “right now there is not a lot of patience to listen, to see the differences between the companies.” Asked whose fault the perception is — his messaging or the market’s — his answer is “I actually don’t think it’s matter”: deliver quarters, and “the investors will figure it out.” Q3 evidence he points to: enterprise wins including Visa and huge healthcare companies far outside tech.

3. The only competitor that scares him is the one he can’t see

• Harry’s challenge: a private Ramp can burn on Super Bowl ads and negative margins while public Navan is scrutinized per dollar. Cohen’s answer is a graveyard story: the year Navan raised $4M, UpSide Travel raised $200M with Priceline’s founder at the helm — “somebody that needs to win against us, that’s them… And they are not around.” On Brex: “I love the Brex guys… but I don’t think that you want to end your journey in Capital One.”
• He rejects the private-market advantage outright: hiding where “nobody will look at how I’m defining revenue, or what’s gross margins… doesn’t create a good business.” The moat is culture — “do you have what it takes to see it through?”
• The paranoia is reserved for unknowns: “the ones like Navan 10, 11 years ago that are ignorant enough about the problem, but are really, really, really good.” His formative scare — seeing Altman demo ChatGPT pre-release at a Napa conference ~4 years ago and telling Ilan, Nina, and Michael: “If we will not build our own platform right now, like right now, we are so dead.”

4. The Salesforce fight: distribution vs. users who hate you

• Cohen repeats his old call that workflow software like Salesforce is “not relevant… not just because of AI, they’re not relevant because that’s not how people want to consume stuff.” Harry disagrees entirely: the only thing that matters is distribution — an AI sales rep that “makes you money while you sleep” sold through Salesforce’s reach “from Chattanooga to Chad” makes it a screaming buy.
• Cohen’s rebuttal rests on one test — “is the user that is using me happy?” — and one line: “I’ve never met a salesperson that told me that they like to use Salesforce.” He lived the distribution thesis at HP software (“it’s like Walmart to sell servers”) and argues that users’ dislike will eventually enable disruption. His concession: distribution matters, “but you can build it.”
• The receipts for user-love-as-moat: “In the history of Navan, we lost six enterprise customers. Five of them are back.”

5. Ava, Navan Edge, and why vertical AI can’t be bought off the shelf

• The stress test as told: the weekend New York’s airports shut down — JFK closed Saturday evening through Monday noon — Ava handled 55% of chats on the hardest support case in existence (“my flight got canceled, the airport was shut down, I’m in panic”), and worst-case call-center wait was 16 minutes. “Try to call any other provider in this space during that time.”
• Why not Sierra or Decagon? “There is nothing even remotely close to what is needed when you go to a complex vertical” — and that includes raw Anthropic/OpenAI APIs, because the tolerance for hallucination is zero: misdirect a traveler from gate B32 and “you’re gonna also get a lawsuit.” Navan uses all the frontier models plus open-source models plus its own data, with an agentic platform routing to the right internal API. The LLM travel-booking demos? “Really nice, cute demos. Trust me, it’s not even remotely close to being there.”
• Next catalyst, as hedged by him: Navan Edge, unreleased, “the biggest or most important release that we’ve ever released” — a new way to buy stuff as a traveler. “Eventually the market will figure out… who has the lead in AI for frequent travelers.”
• On sizing the AI shift, his Amazon-in-2000 analogy: nobody looking at a bookstore predicted AWS. Valuations “only God knows how you will justify them” — yet people are “probably underestimating” the change’s size. Jobs redistribution? “It’s just impossible to know these things.”

6. LLMs: commoditized infra or the opposite? And the six-hour expense product

• Cohen dismisses the LLM-infrastructure debate as “not an important discussion… a very commoditized kind of thing.” Harry’s pushback — Harvey’s and Lagora’s founders both went “from diehard OpenAI to diehard Anthropic” on Opus 4.5, which “suggests the opposite of commoditization.” Cohen’s disarming reply: “I’ll tell you why I’m actually happy that I’m not an investor… I actually don’t have a clue” how you’d choose between them.
• What Navan actually does: Cognition switches between models by answer quality, and many Edge transactions are recently with Google — Edge uses Navan’s own model plus Google. Meanwhile Anthropic’s coding infrastructure is “a massive, massive change” making him rethink engineering itself, “similar to what Microsoft did in the ’90s.” Today’s lead: Anthropic — “but God knows.”
• The moat-redrawing anecdote: Cohen challenged the team to vibe-code Navan’s expense product; Ilan did it in six hours over a weekend (on Cognition, probably — “I’m actually not the right person to tell you which platform he used”). ERP integrations to NetSuite? “They are not [hard]. They are not.” The durable part is travel’s plumbing: licensed and incorporated everywhere, ten years of infrastructure to buy, change, and re-credit tickets — “almost impossible to disrupt us.”
• Downstream, Cohen wonders whether product managers may become the most important people — the product-vs-engineering tension may be going away “by the product guy just saying, I’m gonna do this by myself and prove you wrong. I don’t think that we are there yet.” Cycles shortening he states flat: “I’m certain about that.” Productivity gains: Ilan “will give you a really big number here, which we decided not to share yet,” but engineering investment now flows mainly to AI projects. Harry’s margin worry — inference costs eat you unless users pay more — gets the direct-airline-connection answer: Navan took a revenue hit connecting direct instead of via aggregators, gained data (“this flight is likely to get delayed,” Delta One vs. Polaris seat maps), and the win-win-win eventually got paid for.

7. Is any public-company CEO happy? The psychology of a halved stock

• Harry quotes Jason Lemkin: “I don’t know one happy public company CEO. Not one.” Cohen’s split verdict: telling the story to new investors is the fun part; the bad part is a mechanism that makes everyone obsess over “this grade that you take every millisecond in the share price” — daily moves nobody can explain. He checks the price morning and market close — “people are asking you, and you need to know” — and shrugs: “I’m not selling, or I cannot sell. My employees cannot sell.”
• Harry’s Vlad evidence that mood tracks price — “dead man walking” when Robinhood was down, “walking on water” at $100B. Cohen, who rode Robinhood’s IPO crash as an investor, flips it: the downturn was when Robinhood was “actually defining the company” — cost structure, new products. His own template is COVID, when everyone said nobody would travel again: “an opportunity to think about how to build a better business.”
• Does share price correlate to morale? “One hundred percent.” Employees refresh the app all day; his job is the Bezos-letter move — point to the business and the two-, three-, ten-year frame. Same test as the talent war: when employees left for Zoom at its peak, he asked, “do you think that more people than the people that live on this planet will use Zoom?” Some came back; the ones who chase “the next shiny thing” weren’t culture-aligned anyway.

8. The career mistake and the price he stopped paying

• His biggest reflected mistake: doing another startup in the middle instead of starting Navan earlier — Concur “looked so scary that probably we’ve delayed this journey by two to three years,” and Navan “would be significantly bigger” today. He half-defends the detour: the enterprise experience (HP and others) was needed.
• The rawest exchange is on parenting: five or six years ago he didn’t invest the time; for the last three he has, and “there is nothing more rewarding” — 18-year-old twins, a 12-year-old. Harry corners him: could he have built Navan without paying that price? “It would have. It would, and that’s the price.” So does he regret it? “It’s probably something to figure out with my therapist.”
• His decade-out optimism runs opposite to his day job: AI-driven efficiency pushing people toward “fun, spirituality, experiences” — like his once-a-year solo scuba walkabouts. And the founding myth, per his wife: “the entire Navan was created because of my anxiety in an airport — and it’s so, so true.”