OpenAI's TBPN Mistake, SpaceX’s $2 Trillion IPO?, Iran Disables Amazon Infrastructure
TL;DR
OpenAI’s TBPN acquisition looks more like a vanity/media-control play than a growth engine — Alex Kantrowitz and Liz Hoffman argue that buying a tech-friendly show with roughly 70,000 viewers per episode won’t solve AI’s public trust problem, especially once TBPN loses any claim to independence by becoming part of OpenAI.
The real AI perception gap is with non-users, not Silicon Valley insiders — Alex cites YouGov-style polling showing the most negative views come from people who haven’t used AI or only seen others use it, while regular users are much less negative, which makes TBPN the wrong channel if OpenAI wants to change mainstream opinion.
SpaceX’s rumored IPO would be a market-moving event, not just another tech listing — At a reported $2 trillion valuation and a possible $40-80 billion raise, Hoffman says it would dwarf the average IPO and could force real capital rotation across markets, with knock-on effects even in places like Treasuries.
AI may create blue-collar demand while hollowing out white-collar entry paths — Drawing on Larry Fink’s comments and her own reporting, Hoffman says jobs like welders, electricians, and HVAC workers are in short supply for data-center buildouts, while junior roles in consulting, banking, law, and software face the opposite problem: AI can already do a lot of the drudge work.
Iran’s strikes on Amazon infrastructure in Bahrain and Dubai expose the fragility of the Gulf AI buildout — Alex reports that some AWS availability zones are effectively hard down and being deprioritized internally, while Hoffman notes war-insurance costs have jumped 1,900% and available coverage per asset has plunged from $3 billion to $100 million.
Private credit’s AI exposure is starting to look like a slow-motion bank run, not yet a credit collapse — Hoffman says Blue Owl’s technology-focused funds are seeing redemption requests of 22% and 41%, far above the 5% quarterly gates, but she frames the immediate problem as liquidity panic in illiquid funds rather than proof that all the underlying loans are already bad.
The Breakdown
OpenAI buys TBPN — and nobody on the show really buys the rationale
Alex opens with the week’s strangest headline: OpenAI buying TBPN, the tech talk show with around 70,000 viewers per episode and a real foothold among Silicon Valley execs. Liz Hoffman describes TBPN as clearly pro-industry — light-touch interviews, friendly energy, big-name guests like Mark Zuckerberg, Satya Nadella, and Sam Altman — and says the move feels less like journalism and more like the classic billionaire instinct to own a megaphone.
Why the “distribution story” still feels wrong
Alex lays out the strongest bull case he’s heard: OpenAI spent a tiny fraction of its multibillion-dollar revenue to “own the living room” of founders and decision-makers choosing what AI stack to build on. But both hosts land in the same place: if OpenAI wants enterprise distribution, TBPN is too niche and too inside-baseball, and once a covered company owns the outlet, the credibility discount is immediate.
AI has an image problem — just not with TBPN’s audience
The deeper issue, Alex says, is that OpenAI is solving the wrong communications problem. He cites polling showing the people most negative on AI are the ones who haven’t used it, while regular users are far less hostile — meaning TBPN mostly reaches the crowd already bought in. Liz’s explanation for the backlash is blunt: AI feels powerful, vaguely driverless, built by rich, cartoonish people the average American doesn’t trust, then dropped into communities as giant data centers that seem to consume water and power.
OpenAI’s internal churn and the “no side quests” context
As they record, news breaks that OpenAI executive Fidji Simo is taking medical leave, while other leadership responsibilities are being reshuffled again. Alex notes that despite the online dunking, buying TBPN isn’t really a violation of Simo’s “no side quests” mantra, because that warning was about wasting compute on non-core projects like expensive video generation, not about every distracting business decision.
The job market split: electricians wanted, junior analysts not so much
Liz brings in her Larry Fink interview and says AI’s labor effects are bifurcating fast. Data-center construction needs welders, electricians, plumbers, and HVAC workers right now, but white-collar pyramid businesses — consulting, law, banking, software — are at risk because AI can already handle the spreadsheet-and-PowerPoint grunt work that used to train junior hires. Her image is memorable: the old hiring pyramid narrows until it becomes a chimney, maybe even a diamond, with fewer juniors and more mid-level managers overseeing agents.
SpaceX’s IPO could suck real money out of the market
The second big story is SpaceX reportedly preparing an IPO at a $2 trillion valuation, with a raise that could hit $40-80 billion. Hoffman says that kind of deal is so huge it stops being just a company story and becomes a market plumbing story — a giant sucking sound as bankers scramble to find capital. She also floats a very Musk-specific theory: taking SpaceX public could make it easier down the line to merge it with Tesla, continuing the consolidation of Musk’s empire.
Iran hits Amazon infrastructure, and the Gulf suddenly looks less “engineered safe”
Alex then reports that Iranian strikes have hit Amazon infrastructure in Bahrain and Dubai hard enough that some AWS regions are impaired, with internal guidance telling teams not to expect normal redundancy and to minimize footprint there. Liz says the business world may have lulled itself into treating the Gulf as a polished global hub instead of “a nice house in a bad neighborhood,” and adds a telling data point from the insurance world: war coverage prices are up 1,900%, while available coverage has collapsed.
Private credit, ARR loans, and “we are all funding our own demise”
They close on Blue Owl and the private-credit jitters around AI infrastructure. Liz says the immediate issue is a liquidity mismatch — retail-friendly funds offering quarterly redemptions against illiquid loans — but there’s also a real question around the quality of 2021-2022 vintage software loans, including ARR-based lending that assumed growth would eventually turn into profit. The kicker, as Alex puts it, is almost too neat: capital flowing into AI infrastructure may be helping build the technology that undercuts the software companies sitting inside private equity and private credit portfolios.