HBO’s Deadwood is fixated on gold.
Which makes sense—set in the 1870s in a lawless South Dakota town, it’s a coarsely poetic Western TV show that’s all about the gold rush.
But here’s the thing: For all the show’s talk of gold, the characters (and the audience) almost never see any. I’ve watched all three seasons twice, and don’t remember seeing more than one or two gold bars ever cross the screen.
Such is the nature of a gold rush—risk-taking, entrepreneurial people all sprinting to stake their claim and turn that claim, with mining and experience, into a fortune. But definitionally, few ever see life-changing gold. It’s a corollary I’ve been applying to AI and got to chatting about with Rob Biederman, Asymmetric Capital managing partner.
“Most people looking for gold lost money, and most people backing AI concepts are going to lose money,” said Biederman. “The gains are going to accrue to a relatively small number of firms…Not to be defeatist, but I think in most of these fairly obvious AI spaces, fully 80% or 90% of people who invest are going to lose money.”
Our conversation occurred as the VC sector sifted through a heap of new data. Crunchbase and PitchBook each delivered their latest reports, and the numbers underscored that while the AI boom booms, venture as a whole remains characterized by liquidity limitations and lingering uncertainties.
PitchBook isn’t ready to call a comeback, but deal counts are looking up: In Q2 this year, U.S. VCs invested $55.6 billion in 4,226 deals, marking the highest quarterly deal count since Q2 2022.
Deal value is also at an eight-quarter high, but with a caveat: That high is propped up by heavy hitters and superstars, like when NBA star Luka Doncic scored 73 against the Atlanta Hawks last season. Sure, a win’s a win, but the final score doesn’t reflect a balanced team. That paradigm applies here too: CoreWeave’s whopping $8.6 billion Series C and Elon Musk-led xAI’s $6 billion Series B, which alone account for 26.3% of Q2’s aggregate deal value, according to PitchBook’s analysis. Globally in Q2, AI startups raised $24 billion, making AI the top sector for funding for the first time since November 2022, when OpenAI’s ChatGPT launched.
The uptick in VC funding may be an encouraging sign to many in the industry, but Biederman sees trouble in the numbers: Too many investors are backing too many companies that are just too similar, and that means the returns just won’t be there for LPs.
"My concern, and why I'm speaking out now in some sense, is that as the returns from some AI investments come back and are disappointing, I don’t want people to use that to write off venture capital,” said Biederman. “It just means that perhaps we don’t need to fund 50 or 60 versions of the same idea.”
Biederman’s point isn’t that there’s no gold in the AI hills—he’s very clear with me that he’s not an AI bear. His philosophy is summed up by a sign in Asymmetric’s office that reads “conventional methods, conventional outcomes.” Applied to AI, that means looking for startups in places that aren’t obvious and that might even seem incredibly specific. Think: vertical market software applications that are "10x better" because of AI, where the technology meaningfully improves processes for customers in the legal, healthcare, and insurance industries.
Other use cases he’s excited about include AI applications that evaluate long-term disability and workers comp insurance claims, and use cases where AI is able to turn physical blueprints into instructions for, say, airplane assembly.
“The history of financial bubbles and hype usually shows bad to average returns in obvious spaces and shockingly good returns just off the beaten path,” Biederman added via email.
Biederman’s thesis made me think about the broader generational shift in venture right now, as some funds look to change hands while new ones spin out.
"Gold confers power,” says Deadwood’s arch-villain, George Hearst. “Power comes to any man who has the color.” (That color, naturally, is gold.) And he’s right: Success is transformative, and success in this AI boom could ultimately tip the Silicon Valley scales of power—especially for younger or smaller firms that, on land less traveled, get to see gold.
Weekend reading…In this week’s “Ask Andy” column, Bonobos cofounder Andy Dunn answers a rather philosophical one: What’s the biggest challenge founders are facing today that can't be solved with or by technology? Dunn’s answer is multi-layered: “My first gut reaction was to say ‘fundraising.’ The magnetic, mission-driven zeal that is bedrock to raising venture capital seems to be a hard thing to imagine even generative AI (whatever that is) being able to replicate…Then I thought about your question some more.” Read the whole column here.
See you Monday,
Allie Garfinkle
Twitter: @agarfinks
Email: alexandra.garfinkle@fortune.com
Submit a deal for the Term Sheet newsletter here.
Joe Abrams curated the deals section of today’s newsletter.
