The Learning Corner by Precursor

This week, Mia and Charles dig into Jacob Sansbury's post on how Spawn, a ten person startup, spends 30x its payroll on AI tokens and ships 77,000 changes a month with a permanent fleet of AI agents. Next, they unpack Kyle Harrison's "The McDonaldization of Venture Capital," which argues that with AI taking 82.7% of US venture dollars and the biggest firms soaking up most of the capital, venture is optimizing for more burgers instead of better ones. They close with a sharp note from Chris Tottman on the founder refrain "investors didn't understand our business," and how to tell whether that's true or a sign you've forgotten what an outsider needs explained. Together, the three pieces ask the same question from different angles: when the machines and the money both pull toward the average, where does the truly new idea come from?

Charles and Mia break down Anthropic's S1 filing, from $16.3B in first half revenue to Charles's question of whether San Francisco is living in the future or on a "feature branch" the rest of the world never adopts. Next, Nick Rubin's dinner recap on the gap between GPUs and usable compute opens a conversation on capital as the new moat, the credit markets financing the AI buildout, and growing local opposition to data centers. Finally, Jake Gibson's "On Risk" argues venture is starting to look like the public markets, and Charles explains why founders now need to think like VCs and why chasing 2021 ARR benchmarks means playing the wrong game. Three pieces, one thread: AI is minting enormous wealth for a small few, and the rest of the ecosystem is still figuring out what that means for them.

For our 100th episode, we turn the mic on Charles to reflect on the state of venture, using the themes that have come up again and again across roughly 300 good reads since we started the show. We dig into whether the skills that made investors great still matter in an AI-first market, drawing on pieces like Alberto Romero's "You Spent Your Whole Life Getting Good at the Wrong Thing" and Sarah Guo's "Relevance Decays." Charles looks back on what surprised him most over the last two years, from the seed squeeze and the bubble debate to founders who now say they work for their AI agents. We close on where venture needs to go to feel exciting again, and the one line summary of venture in 2026 Charles would leave for someone reading it in 2036.

This week Mia and Charles dig into Kyle Harrison’s “The Death of Venture Capital’s Short Kings,” on Benchmark and USV, the two firms that built their identities on staying small, both raising their largest funds ever this year. Then Nnamdi Iregbulem’s “Valuations Need a Correction” argues that seed valuations are financing prices for the few companies that successfully raise, not a read on what the average startup is worth. Once you correct for that, seed market value is climbing while the number of active seed startups shrinks, a concentration story that looks a lot like the Mag 7 and the S&P 500. Two pieces, one thread: more capital and higher marks are telling a much narrower story than the headlines suggest.

This week Mia and Charles start with Steve Vassallo's argument that founders obsess over cash runway when the clock that actually matters is relevance runway, and that plenty of well funded companies are racing against good enough AI versions of their own product. Then they dig into Venky Ganesan's 20VC episode with Harry Stebbings, where he says every seed check is an option bet and large funds are indifferent to seed valuations because they are buying a seat at the table. Charles answers from the pre-seed seat: is a Precursor check an option or a bet, and does ownership still matter before a company is an outlier?

This week on The Learning Corner, Mia and Charles welcome Rob Go, Cofounder and Partner at NextView Ventures, whose writing has been a staple of our reading list since the beginning. We revisit his piece arguing that seed VC is facing a structural crisis rather than a normal down cycle, and the one question he says the whole debate comes down to: classic venture or super compounder? Rob also breaks down the four approaches to early stage venture and how funds drift down the valuation curve without meaning to.

This week on The Learning Corner, Mia and Charles dig into Andreessen Horowitz's $35 million bet on the Horowitz Andreessen Academy, a two year college alternative backed by Anthropic, Google, Meta, Nvidia and OpenAI. They then unpack a sharp post from signüll arguing this may be both the best and worst time to start a company, as AI makes building cheap and advantages disappear just as fast, with the big labs close behind. The episode closes with Venky Ganesan of Menlo Ventures on reflexivity in AI valuations, house money versus FOMO, and why position sizing may matter as much as picking winners.

This week on The Learning Corner, Mia and Charles dig into Gil Dibner's argument that venture has barbelled into two extremes, a tiny set of consensus darlings that can raise at any price and everyone else, who are increasingly paying pure gold prices for fool's gold. Then they turn to Dario Amodei's essay "We Must Pace the Frontier," the rare moment where Amodei, Sam Altman, and Elon Musk all agreed the AI industry needs to slow down. They unpack what "pacing" actually means, and whether it is good or bad news for the thousands of startups building on top of these models.

This week Mia and Charles dig into a market that only seems to have two speeds. Ethan Kurzweil of Chemistry argues the trillion dollar startup is a pink elephant nobody can unsee, and that the pedestrian double or triple has become a relic of a bygone era. David Cahn looks at what happens after the money lands: a generation of well funded teams with big visions and no clear steps to get there. Nikunj Kothari of FPV Ventures closes it out with a tactical playbook for founders raising this fall, from the number you say out loud to why VCs don't actually want a "deal."

This week we sit down with Nic Poulos, Co-Founder and General Partner at Euclid Ventures and one of the sharper voices on vertical AI, to pressure test the thesis as we head into the end of Q3. We dig into what defensibility actually looks like when any founder can spin up a vertical AI product in a weekend, why a $500M median exit is great math for some funds and tough math for others, and where "death of SaaS" narrative currently sits. Nic also revisits his 2026 predictions on deep tech fragmenting and religion as a vertical AI opportunity.