Latest / Tech IPO Conversations with Fexingo: Public Software Companies and Capital Markets / Inflated ARR Metrics Are Warping AI Startup Valuations
Transcript
- Lucas: There was a TechCrunch piece this week that really got under my skin — in a good way — about how VCs and founders are using inflated 'ARR' numbers to crown AI startups. And I think if you're paying attention to the IPO pipeline, this is the single most important thing to understand right now. Luna: Define 'inflated' here. Are we talking about outright fraud, or creative accounting? Lucas: More the latter, but with real consequences. The article — by a reporter who's been covering SaaS metrics for years — walks through several cases where startups are counting things like pilot contracts, non-recurring professional services fees, and even future commitments as 'annual recurring revenue.' The number gets stretched to the point where it's misleading anyone who doesn't ask the right questions. Luna: And this matters especially for AI companies right now because...? Lucas: Because the hype cycle is at its peak. We're seeing valuations that assume these companies will grow into massive recurring revenue businesses — but if the underlying ARR is inflated, the math breaks. Look at Palantir up 1.3% this week, Coinbase down 2.3% — the market is still rewarding some of these names, but the bar for profitability is getting higher. Luna: Right. And we've seen this movie before — pre-2021 SPAC era, where companies were touting 'projected revenue' that never materialized. Lucas: Exactly. The article highlights one AI startup that claimed $50 million in ARR. When you dig in, $20 million of that came from a single pilot contract with a large enterprise — a six-month trial with no renewal guarantee. Another chunk was professional services to customize the model, which is inherently one-time. Luna: So the real recurring piece might be half of what they're presenting. That's a massive difference for anyone doing valuation work. Lucas: Huge. If you're pricing a company at 20 times ARR — which is common for high-growth AI — and the ARR is inflated by 40 percent, you're overpaying by 40 percent. That's not a rounding error. Luna: It reminds me of the 'adjusted EBITDA' games we saw in the 2010s. Companies would add back every possible cost and call it profitable. Lucas: Exactly. And the SEC has started paying attention to ARR definitions in S-1 filings. I was looking at a recent IPO prospectus from a mid-stage software company — they had a whole section defining what they count as ARR, and they explicitly excluded professional services and pilots. That's the gold standard. Luna: But private companies don't have to file with the SEC. So investors doing due diligence have to ask the right questions themselves. Lucas: And that's where it gets tricky. A lot of late-stage investors are under pressure to deploy capital. They see a hot AI company with fast-growing ARR and they jump in without verifying the composition. Luna: What are the specific questions they should ask? Lucas: The TechCrunch piece has a good checklist. First: what percentage of ARR comes from customers with less than twelve months of history? High churn risk there. Second: how much is from pilots versus long-term contracts? Third: are there any non-recurring revenue streams included — like setup fees or consulting? Luna: And what about the 'future commitments' angle? I've heard of startups counting signed contracts that haven't started generating revenue yet. Lucas: That's a big one. Some companies count the total contract value of a multi-year deal as ARR, even if the revenue is recognized over time. That inflates the number upfront and misrepresents the actual annual run rate. Luna: So for listeners who are following the IPO pipeline — maybe they're thinking about investing in the next big AI IPO — what's the takeaway? Lucas: Don't take ARR at face value. Look for a footnote or a reconciliation. If the company is public, it's in the filings. If it's private, ask the VC or the founder directly. And if they can't or won't give you the breakdown, that's a red flag. Luna: Speaking of supporting the work we do here — if today's episode was useful to you and you want to keep it ad-free, buy me a coffee dot com slash fexingo helps. Lucas: Yeah, really does. A handful of listeners chip in monthly through that link — it literally funds these conversations. Anyway, back to the ARR question. Luna: So where do we see the fallout first? Will some of these AI startups hit a wall when they try to go public and the underwriters start sanity-checking the numbers? Lucas: I think that's exactly what will happen. We've already seen a few IPO filings get delayed or pulled because the company's growth narrative didn't hold up under scrutiny. The market is more skeptical now than it was two years ago — partly because of the SPAC hangover, partly because interest rates are still relatively high. Luna: And the data this week shows that. The S&P 500 is flat-ish, but the ARK Innovation ETF is up 2.5% — so there's still appetite for growth stories, but it's selective. Lucas: Right. The market is rewarding companies that can demonstrate real, recurring revenue growth with healthy gross margins. But the window for 'story stocks' is much narrower. Luna: Let's talk about a specific case from the article — the AI startup that had $50 million in ARR but only $30 million real recurring. What happened to them? Lucas: They were in the process of raising a Series C at a $1 billion valuation — a unicorn round. When a potential lead investor did the diligence and found the inflated ARR, they walked away. The round got downsized, the valuation dropped to $600 million, and the company had to restructure. Luna: Ouch. So that's the real-world impact of this metric abuse. Lucas: Yes. And it's not just about the founders getting a lower valuation — it's about the employees who have options priced at that higher number, and the earlier investors who thought they were sitting on a paper gain. When the truth comes out, everyone takes a hit. Luna: Is there any regulation that could help? Should the SEC step in and define ARR more strictly for private companies? Lucas: The SEC has limited authority over private companies that aren't raising money from the general public. But there are moves to tighten the rules around 'general solicitation' and the kinds of disclosures that are required in private placements. That could indirectly force more rigorous ARR definitions. Luna: In the meantime, it's caveat emptor. Lucas: Absolutely. And the best defense is education. Which is why I wanted to spend time on this today. Luna: It's a good reminder that even in a hype cycle, the fundamentals still matter. And ARR — when defined correctly — is one of the most powerful metrics for SaaS and AI companies. But only if it's honest. Lucas: Right. A well-defined ARR number gives you a real sense of the company's recurring revenue base, its growth trajectory, and its retention. But when it's stretched, it becomes a weapon for hype rather than a tool for analysis. Luna: So next time you hear an AI startup touting its ARR, ask: is that real recurring revenue, or is there something else hiding in there? Lucas: Exactly. That one question could save you a lot of money.