Transcript
SPEAKER_1: Let's start with something that really stands out here. Britain's AISI published a major capability finding this week, and the contrast with the US is striking. What's the institutional reality there? SPEAKER_2: The US simply has no equivalent body to Britain's AISI. While the UK is publishing actual capability findings, the US government is debating org charts. That's the gap we're talking about. SPEAKER_1: And in practical terms, what does that mean for where the models actually are right now? SPEAKER_2: Mythos has already cleared the cyber ranges. GPT-5.5 is one range behind. The next Google Gemini model drops at I/O next week. Every frontier model company is shipping offensive capabilities faster than any government can evaluate them. SPEAKER_1: So the ONCD and CAISI can argue about turf, but the models aren't waiting. SPEAKER_2: Exactly. That's the bottom line. The bureaucratic debate is happening in one lane, and the technology is racing ahead in another. SPEAKER_1: Now let's get into the Musk versus Altman trial, because closing arguments wrapped this week. What did the trial actually reveal? SPEAKER_2: What it revealed about the OpenAI-Microsoft relationship is almost as interesting as the case itself. Court filings showed Sam Altman holds over two billion dollars in stakes in companies that did business with OpenAI, including a one point seven billion dollar position in fusion startup Helion Energy. SPEAKER_1: And Helion signed a power purchase agreement with OpenAI, right? So Musk's team is using that to argue Altman steered resources toward his own interests. SPEAKER_2: That's the argument. And on the Microsoft side, executive Michael Wetter testified that Microsoft has spent over one hundred billion dollars on OpenAI, combining equity investments and infrastructure buildout. SPEAKER_1: One hundred billion dollars. And there's also a revenue-sharing cap that came out of this? SPEAKER_2: Yes. OpenAI and Microsoft have capped their revenue-sharing arrangement at thirty-eight billion dollars, which clears the path for OpenAI's full for-profit conversion. Internal emails also surfaced showing Microsoft feared being too dependent on OpenAI. SPEAKER_1: So what's Altman's side of the story in all this? SPEAKER_2: Altman's team argued Musk abandoned OpenAI after losing a control dispute and is now interfering because xAI is a direct competitor. OpenAI's lead researcher also testified that at a twenty seventeen meeting, Musk pushed to race toward AGI to beat DeepMind, while safety staff objected. SPEAKER_1: So it's really two different visions for the same nonprofit, and a bitter split when Musk lost that control dispute. SPEAKER_2: Right. And at its core, this isn't really about the legal technicalities of a nonprofit conversion. It's about who gets to write the origin story of the most important technology company of the decade. SPEAKER_1: Whether Altman is a self-dealing founder or a visionary who made hard calls under pressure. SPEAKER_2: Exactly. And the judge decides alone, no jury. The relief Musk is seeking includes blocking OpenAI's for-profit conversion, unwinding Microsoft's influence, and a range of financial damages. SPEAKER_1: Walk me through what each outcome actually means. SPEAKER_2: A ruling for Musk means OpenAI's cap table gets messy, the nonprofit conversion stalls, and the one hundred fifty billion dollar valuation is at risk while investors wait for clarity. SPEAKER_2: A ruling for Altman means OpenAI moves full speed into Deployment Company mode, Microsoft's one hundred billion dollar bet pays off, and xAI is positioned as the scrappy underdog fighting the monopolist. SPEAKER_1: Either way, the AI power structure shifts. And regardless of the verdict, a lot is already on the record. SPEAKER_2: That's the key point. The OpenAI-Microsoft relationship is now a publicly documented one hundred billion dollar bet. Altman's personal stakes are public record. The broader governance question, who actually controls the most powerful AI lab in the world and who benefits, is no longer hypothetical. SPEAKER_1: Let's shift to the money side of things, because three separate clusters of earnings and deals landed this week. What's the bigger picture they paint together? SPEAKER_2: Taken together, they describe something bigger than any individual headline. The AI infrastructure economy is compounding. SPEAKER_1: Start with Cisco, because those numbers were pretty remarkable. SPEAKER_2: Cisco posted record Q3 revenue of fifteen point eight four billion dollars, up twelve percent year over year, and raised its fiscal year twenty twenty-six AI infrastructure order forecast from five billion to nine billion. The stock jumped over fifteen percent. SPEAKER_1: And they cut nearly four thousand jobs to fund that AI pivot. SPEAKER_2: Yes. Hyperscaler networking orders grew fifty percent year over year. Cisco is no longer a legacy networking play. It's now one of the largest AI infrastructure vendors by order volume. SPEAKER_1: And then there's Foxconn, which is a fascinating story in itself. SPEAKER_2: Foxconn reported Q1 revenue of sixty-seven billion dollars, up twenty-nine percent year over year, with AI servers now its largest revenue segment. The CEO said Q2 will show significant growth versus Q1 on robust AI demand. The company that makes your iPhone is now primarily in the business of making AI compute. SPEAKER_1: There was also something interesting involving Jensen Huang's charitable foundation this week. SPEAKER_2: His foundation purchased one hundred eight point three million dollars in CoreWeave compute time and donated it to universities. It's a strategic philanthropic move that puts CoreWeave hardware in front of the next generation of AI researchers. SPEAKER_1: Now let's talk about the IPO wave, because Cerebras was a big one. SPEAKER_2: Cerebras priced at one hundred eighty-five dollars per share, raising five point five five billion dollars at a fifty-six point four billion dollar fully diluted valuation, above its expected range of one hundred fifty to one hundred sixty dollars. The deal was twenty times oversubscribed, making it the largest US tech IPO since Snowflake. SPEAKER_1: But there's a catch, right? A significant concentration risk. SPEAKER_2: Sixty-two percent of revenue comes from a single UAE government customer. One sharp analyst put it well: public AI exposure is so scarce relative to demand that the market will pay through the concentration risk. SPEAKER_1: And Blackstone also went public this week with a data center play. SPEAKER_2: Blackstone Digital Infrastructure Trust raised one point seven five billion dollars in its US IPO, selling eighty-seven point five million shares at twenty dollars each, and plans to acquire data centers in the two hundred fifty million to one point five billion dollar range. The picks-and-shovels AI trade is going public in bulk. SPEAKER_1: And then there's the big tech restructuring happening at the same time. Meta is making some significant moves. SPEAKER_2: Meta is cutting approximately ten percent of its workforce on May twentieth and has already moved over one thousand engineers into a centralized AI unit to build generative models. WIRED spoke with sixteen current and former employees, and the word used most often was miserable. SPEAKER_1: And reportedly, anyone who can afford to leave is hoping to be laid off just for the severance package. SPEAKER_2: That's what was reported, yes. It paints a pretty stark picture of the internal culture right now. SPEAKER_1: Microsoft is also making moves, including a potential acquisition. SPEAKER_2: Microsoft is in discussions to acquire LLM developer Inception for over one billion dollars, described as preparation for a future independent of OpenAI. SpaceX also pursued Inception. And at the same time, the UK CMA launched a formal antitrust investigation into Microsoft's business software dominance, targeting the bundling of Windows, Word, Teams, and AI integrations. SPEAKER_1: So Microsoft is simultaneously expanding and being scrutinized at the same speed. SPEAKER_2: That's exactly the situation. It's a company moving aggressively on multiple fronts while regulators are starting to catch up. SPEAKER_1: To bring it all together: the US lacks an institutional body to match Britain's AISI while frontier models outpace government evaluation; the Musk-Altman trial has put OpenAI's governance, Altman's financial conflicts, and Microsoft's one hundred billion dollar bet all on public record; and the AI infrastructure economy is compounding fast, from Cisco and Foxconn earnings to the Cerebras IPO and Meta and Microsoft restructuring for an AI-first future. Next time, we'll be wrapping up the course with lecture three of three.