Benchmark’s Ship of Theseus, OpenAI Kills Sora, SpaceX $2T IPO Buzz | Diet TBPN
TL;DR
TBPN asks whether Benchmark can ever be forgiven — and frames it as a Ship of Theseus problem — only 2 of the 6 equal GPs from the 2017 Uber ouster era remain (Peter Fenton and Eric Vishria), raising the question of whether a partnership becomes a different firm as partners turn over.
Their core claim is that Benchmark may have cost Uber hundreds of billions by pushing out Travis Kalanick and abandoning in-house autonomy — with Uber at roughly $150 billion, Waymo at $126 billion, and self-driving now the main valuation overhang, they argue a Travis-led Uber could plausibly have been a $500 billion to $1 trillion company.
The steelman of Benchmark isn’t moral courage, it’s panic over concentrated wealth — the hosts suggest each partner was staring at a potential $1 billion outcome on one power-law deal, saw boycotts, hit pieces, and Lyft competition piling up, and chose to de-risk the asset rather than bet on Travis through the storm.
OpenAI killing the standalone Sora app is presented as a good sign of product discipline, not a retreat from video — Tyler Hogge’s take is that “bullish killing products quickly is hard,” and the likely move is consolidation into ChatGPT rather than abandoning AI video entirely.
AI video hasn’t become the infinite-content doom loop people feared because distribution and compute still dominate — Sora and Google Veo-style tools were cool creative products, but creators still post finished MP4s to TikTok or Instagram, and brutal rate limits on $250–$500/month plans wreck retention before a native network can form.
Fiverr’s giant ‘Find the best AI directors’ billboard shows both the opportunity and the trap for labor marketplaces in the AI era — AI is eating the low-end freelance tasks Fiverr was built on, so the company is pivoting toward professionals who orchestrate tools like Runway, Higgsfield, and Chinese video models, but the old disintermediation problem remains if clients just hire creators like Billy Bowman directly.
The Breakdown
Benchmark, Travis, and the decade-late forgiveness question
The show opens in deliberately dangerous territory: can Benchmark ever be forgiven for helping force Travis Kalanick out of Uber in 2017? The hosts say everyone agrees the fallout was ugly, but after a decade and major partnership turnover, they ask whether a venture firm is still the “same” firm when most of the people have changed.
The Uber counterfactual: what if Travis had stayed?
They lean hard into the alternate-history case that Uber under Travis could be worth far more today — Jason Lemkin’s reaction to their Travis interview was that it’s hard to imagine Uber being worth less than $1 trillion. Uber survived and sits around a $150 billion market cap, but they argue that’s basically a honed, managed version of the machine Travis built, not some radical reinvention.
Self-driving is the missing piece — and Waymo makes the loss feel bigger
The hosts say the real valuation drag on Uber now is autonomy: Dara has answered the self-driving question with partnerships and investments, not internal IP. With Waymo valued around $126 billion in February and Uber at $150 billion, they argue you can almost see the missing market cap sitting there — plus whatever “Travis premium” public markets would have assigned to a one-of-one founder still in the seat.
Benchmark as a venture Ship of Theseus
This is where the Greek mythology comes in. In 2017, Benchmark’s equal GP roster included Bill Gurley, Eric Vishria, Matt Cohler, Mitch Lasky, Peter Fenton, and Sarah Tavel; today only Peter and Eric remain, with Chetan Puttagunta, Everett Randle, and Jack Altman joining later. So the question becomes: if only one-third of the original partnership is left, and newer partners like Everett were literally in college during the Uber saga, when does the brand become a new ship?
The “steelman” defense: they weren’t noble, they were terrified
The hosts then offer what they call the full steel-helmet steelman: each Benchmark partner was staring at a clean $1 billion payday from Uber, likely dwarfing everything else in their portfolio. As boycotts, Mike Isaac stories, movies, and Lyft pressure stacked up, they suggest the firm panicked — not because of a pure moral stand, but because 99% of perceived net worth seemed suddenly at risk.
Sora dies as a standalone app, and that’s probably healthy
From there the show pivots to OpenAI sunsetting Sora’s app. Their read is not “Sora failed,” but that OpenAI is finally consolidating after its recent “code red” and refocus chatter: the best version of video generation probably lives inside ChatGPT, because even fans of Sora didn’t enjoy having to bounce into a separate product.
Why AI video is real but not yet a network
They revisit the launch vibes around Sora and Meta’s video efforts: the fear wasn’t that these products would flop, but that they would “one-shot humanity” with an entertainment doom loop. Instead, AI video looks more like a creative tool than a destination, because creators generate an MP4 and immediately post it to TikTok or Instagram, while brutal compute limits — even on Google plans costing $250 to $500 a month — kill the endless-scroll retention that real networks need.
Fiverr’s billboard, AI directors, and the weird middle ground of AI labor
The last big segment riffs on Fiverr’s flashy Hollywood billboard pushing “AI directors,” complete with KTLA coverage and a joke that the sign caused a rear-end collision. The hosts think the campaign is a clever pivot for a company squeezed by AI — low-end freelance tasks are getting one-shotted, but serious AI video still needs someone managing prompts, APIs, fine-tunes, and multiple models — though they also note the classic marketplace problem: if Billy Bowman is great, clients can just hire Billy Bowman directly.