Ramtin Naimi: Inside the 2024 AI Venture Boom

In the bars and Slack channels of San Francisco's venture crowd, there is a joke that isn't really a joke. An investor who loses the chance to lead a hot funding round no longer waits politely for the next one. He pays his way back in immediately, at three to five times the price the winning investor just locked in. The slang for this, coined by the people who keep paying it, is the "cuck tranche" (02:46). It is a crude name for a real mechanism, and it tells you most of what you need to know about where venture capital stands right now.

Ramtin Naimi, co-founder of the venture firm Abstract Ventures, is the guest supplying the view from the deal floor. Host Patrick O'Shaughnessy asked him to skip the usual origin story and zoom in on the present moment, and Naimi obliged with an unusually granular account of how fast, strange, and lucrative Silicon Valley has become. Rounds close in days. Follow-ons happen before the ink on the last round dries. And the firms that lose a deal they wanted no longer wait for the next financing to buy in. They pay up immediately, in parallel with the primary round, at three to four or five times the price (03:01).

"What do we need to pay to get into this financing today? And it's usually not two X. It's usually like three to four to five X happening in tandem with the primary." — Ramtin Naimi [03:01]

Naimi doesn't read this as a symptom of mania. He argues these companies are so capital-hungry that splitting a round into tranches is a necessity, not a red flag (03:20). What has changed is the psychology around it. Getting a markup used to feel like a win. Now, he says, "getting markups is the bare minimum requirement to be keeping up to the point where markups don't even feel like good anymore" (03:57). A founder who isn't marked up every few weeks starts to feel like he's falling behind, even when nothing about his business has changed.

The speed shows up most clearly in how venture firms raise their own money. A fund normally plans an eighteen-to-twenty-four-month window to deploy its initial checks before coming back to investors for more (10:32). Naimi says the largest platform funds are now burning through that money roughly three times faster than modeled (11:14). He gives two anonymized examples: one fund raised two and a half billion dollars in July and told him it would be back in market by January, and another was handed three billion dollars nine months before the taping and is already back asking for five billion (13:49). The gap between the ownership stakes funds want and the prices they actually have to pay to get them is the hidden driver. Everyone underestimated the check sizes required to buy the same slice of the same companies.

That mismatch cascades down to limited partners, the endowments and foundations that fund venture firms in the first place. Naimi says LPs expected far more liquidity by now and are discovering an uncomfortable truth: much of their exposure to venture capital is really exposure to three or four companies, SpaceX, Anthropic, OpenAI, and Databricks among them (12:53). If your fund wasn't in those names, there is no liquidity coming. That leaves every other fund facing a choice described starkly in one of the episode's sharper passages.

"LPs are kind of looking at these guys and realizing that there's two things that are happening. They either maintain their ownership in these companies, but they end up with way fewer shots on goal." — Ramtin Naimi [11:50]

Either a fund protects its ownership stake and ends up placing fewer bets overall, or it keeps the same number of bets and accepts a smaller stake in each one. Pick the first path and you need to be a very good stock picker. Pick the second and you are betting on sheer volume. There is no comfortable middle.

The Revenue Isn't What It Looks Like

Naimi's second warning concerns the quality of the growth everyone is chasing. Companies posting ten, fifteen, twenty times year-over-year growth are real, he says, but almost none of them have clean, ninety-percent-margin, pure subscription revenue (04:49). Data-labeling companies are growing at wild rates while holding margins of just fifteen to twenty percent, because collecting the data itself is expensive (05:59). Companies built on top of foundation models often hand most of their margin straight to the model provider. And yet the trend line is improving fast: Anthropic's gross margins hit eighty-five percent in the first half of the year (06:18), and Naimi says portfolio companies are moving from negative fifteen percent margins to positive thirty-five percent while also multiplying revenue several times over.

This is also why vertical software companies selling into specific industries, like law firms, are commanding roughly a hundred-times revenue multiple while public SaaS companies trade at around thirteen times (06:41). The bet isn't that these are software companies at all. It's that they are quietly becoming labor-replacement businesses, and the market that matters isn't what law firms spend on software, it's what they spend on paralegals (07:14).

Naimi's comparison to 2021 is the most useful lens in the whole conversation. Back then, he says, "the companies weren't that interesting when you looked at them" (07:33); investors were just comparing them to nothing else. Today he personally uses several portfolio products daily, and he's watched family members who never adopted any tech product start relying on AI tools as part of their routine (09:32). What is similar to 2021 is the behavior: term sheets offered and pulled within days, and in the last six to eight weeks before this recording, more renegotiated deals than at any point since 2021 (09:10).

Pricing a Trillion-Dollar Imagination

Underneath all of it sits a shifting idea of what a company can become. Naimi says that when he started investing in 2016, a ten-billion-dollar outcome was the ceiling most seed deals were priced against (24:06). By 2021, with Stripe, Databricks, Airbnb, and DoorDash nearing hundred-billion-dollar valuations, that ceiling moved to a hundred billion. Now, with SpaceX, Anthropic, and OpenAI each approaching or exceeding a trillion dollars, every seed deal gets priced as if a trillion-dollar outcome is the best case (24:54). Naimi doesn't think there will be forty trillion-dollar companies. He thinks there will be two or three more. But the imagination required to write a check has permanently expanded.

His framework for surviving that expanded imagination is a distinction between reactive and proactive investing. Reactive investors chase the deal everyone already wants and, if they pick correctly, can be top-quartile managers. Proactive investors back brilliant founders in categories nobody cares about yet, and that, he argues, is what separates the top five percent (26:36). His example is Conduit, a brain-computer-interface startup Abstract seeded about nine months before this recording, when almost no one in Silicon Valley cared about BCI (26:53). Conduit's wearable device reached eighty percent accuracy predicting what someone was about to type, built on a small initial set of human training data that later grew to thirty thousand hours (27:34). Within eight to ten weeks, BCI became a hot category and Conduit's next round drew thirteen competing term sheets (27:55). Chip startup Etched tells a similar story: nearly every investor who met founder Gavin Uberti thought he was brilliant, but almost no one believed a startup could out-compete incumbents in AI chip fabrication, until Sequoia Capital priced the company at ten billion dollars and the market suddenly agreed (28:55).

Compute itself has become the real currency behind every number in this conversation. Naimi notes that some VC firms now finance AI startups with compute allocations rather than cash, since founders know the money will be spent on chips regardless (17:07). The scale of the underlying spend came into sharp relief the same week Anthropic filed its S-1: the company had raised a hundred and forty billion dollars, burned ten billion of it, and still had a hundred and thirty-two billion dollars on its balance sheet at eighty-five percent gross margins (15:50). That is the backdrop against which Naimi rates the opportunity set a ten out of ten relative to his nearly ten years in venture, and the overall frothiness a six and a half to seven out of ten, lower than 2021's ten, because unlike 2021, he says, there is real adoption behind the hype (43:30). Whether that distinction holds is, as always in venture, a bet nobody can place with certainty until the companies now worth a hundred billion either become worth a trillion or don't.