|
9/30/2026
Hot Tip - Buy Low, Sell High! When we were raising our last fund in 2024, private credit was all investors wanted to talk about. Family offices and RIAs told us they could get close to venture-like returns with much more security. This month U.S. private credit defaults hit a record 6.3%. The warning signs were there early At an investor lunch in 2019, the private credit manager who presented after us said lenders were already relaxing terms and covenants just to get allocations. When everybody thinks something is a good idea, it gets over-bought. There are only so many good loans to go around. The money kept coming anyway. Private credit fundraising hit record levels in the first half of this year. Loan spreads actually narrowed as more lenders chased the same deals, and payment-in-kind loans, where borrowers add unpaid interest to the balance instead of paying cash, climbed to nearly 10% of business development company portfolios (Axios). Now there's so little agreement about the real default rate that estimates run anywhere from 1% to 19% (Bloomberg), and lenders are preparing to take over the companies they lent to. Venture learned the same lesson in 2021 In 2021 a lot of ordinary SaaS companies got priced like something extraordinary. The frenzy was driven by the supply of capital, not the supply of great companies. Investors who got burned at the top understandably wanted something safer. What the numbers say To be fair, venture has struggled in recent years. Cambridge Associates tracks net returns to investors across thousands of U.S. venture funds. As of mid-2025: Period U.S. venture S&P 500 Russell 2000 10 years 13.1% 13.9% 7.2% 15 years 15.3% 15.5% 11.2% 20 years 12.2% 11.0% 8.2% Over the last decade venture roughly matched the S&P 500, which looks even worse when you factor in illiquidity. Over 20 years it has kept pace with large-cap stocks and beaten small caps. For comparison, the Cliffwater Direct Lending Index, a common private credit benchmark, has averaged 9.5% a year over the same 20 years. The vintage matters more than the decade Averages hide the cycle. Returns depend heavily on when a fund starts investing. Using Cambridge Associates' vintage-year data:
Before 9/11, I trained as an Apache pilot to stop hordes of tanks coming from Russia or North Korea. Then we deployed to fight insurgents in cities, which looked nothing like what we'd trained for. Investors who were bucking the trend and investing in private credit in 2020 and 2021 instead of venture probably did very well. Those who continued to pile in later may have been fighting the last war: protecting against the last cycle's risk while crowding into the next one. The boring fix Buy low, sell high is the oldest rule in investing, and humans just aren't wired to do it. The fix is boring: set your allocation and keep it, especially for whatever is out of favor. GPT-3 came out in 2020 and ChatGPT in 2022, years when venture was out of favor. By the time venture looks great again, it might be too late to get in. Sources: Fitch via Private Equity Wire · Axios · Bloomberg · Cambridge Associates 1H 2025 · Cambridge Associates vintage data · Cliffwater This post reflects the author's views and is for informational purposes only. It is not investment advice or an offer to sell any security.  Most venture firms don't make it to fund three. We don't say that to brag — we say it because we think about it. Building a durable investment firm is hard, and reaching this milestone is something we're genuinely grateful for. It happened because of the founders who trusted us early, the LPs who believed in what we were building, and a community in Colorado — and now Texas — that has grown into something none of us could have fully predicted in 2014.
So: thank you! And here's where we go from here.  This month’s Four-Minute FirstMile Founder interview features Patrick Leonard, founder and CEO of SideBy Care, which is leveraging AI and virtual care to treat gut-brain disorders. An experienced digital health product leader and repeat founder, Patrick has built and scaled some of the largest virtual care and healthcare AI platforms in the industry, including CVS Health Virtual Care, iTriage (20M users), and Aetna Digital.
By, Zaz Floreani Carta, PitchBook, KPMG, and Cooley data all point to the same story: fewer Series A rounds, slightly higher prices, and a widening gap between AI-driven growth and everything else. 2025 may be shaping out to be a year of a “two-track” Series A market.   This month’s four-minute FirstMile Founder interview features Lisa Kagan, CEO and co-founder of Buddy, a FirstMile 2024 portfolio company that is building a scrap metal marketplace for the future. Over the past 17 years, Lisa has helped conceive, build, and grow a number of ventures across different industries.
10/14/2025
When AI Becomes the Attack Surface It’s 2025 and AI agents are now critical to many business processes. This is creating a massive shift in security needs. AI is now an attack surface, not just a tool. Traditional defenses fail against attacks on the model's reasoning layer. Layered defenses must now include reasoning-layer protection to detect malicious intent pre-action.
By, Bill Miller 9/25/2025
AGI by 2030: Plausible or Probable?In my fourth post of our AI Series, I share my critical response to Google DeepMind’s “An Approach to Technical AGI Safety and Security” published in April 2025 that famously predicts Artificial General Intelligence (AGI) will be here by 2030. By, Bill Miller  In my first two posts of our AI series, I explored how today’s AI coding assistants bump into memory gaps when context windows are too short, and why code-generation experiments often feel like working with a very smart, but distractible, junior teammate. In this third installment, I look at the flip side: the problems that arise when context windows are too long. The culprit is a phenomenon researchers call “context rot”. By, Bill Miller  |
