AI's Big Week: IPOs, Apple's Siri Makeover, and the Startups Betting Billions on What Comes Next

LaurenMm-hmm.

DerekWelcome back to Tech Insider Weekly. I'm Lauren, and we have a genuinely stacked show today.

Speaker 3Derek here. Stacked is underselling it, honestly.

DerekOkay, so get this. The WSJ is reporting OpenAI confidentially filed to go public.

Speaker 3Wait, for real? Like an actual IPO filing?

DerekAn actual filing. And they turned right around and said it may be a while before anything happens. happens.

Speaker 3So, they filed to tell us they're not ready to file.

DerekPretty much. We're going to dig into what that actually signals and whether the numbers can hold up the hype.

Speaker 3I have thoughts, strong ones.

DerekShocking. Also on the show, Apple WWDC dropped yesterday.

Speaker 3Siri AI, powered by Apple Intelligence, and now apparently pulling in Gemini for world knowledge queries.

DerekAbout that Google partnership, we're going to debate whether this is a real architectural shift or just repackaged promises.

Speaker 3Two years of stumbles is a lot of ground to make up.

DerekIt is. And then we've got a wild VC funding roundup. Cyera hit a $12 billion valuation, up fourfold in 18 months, according to CTech.

Speaker 3Fourfold in 18 months?

DerekYeah, yeah. Plus NinjaOne raised $400 million. And there's this fascinating story TechCrunch has about a guy who deployed nearly 500 million into hot startups without ever running a traditional VC fund.

Speaker 3Okay, that one I need explained to me.

DerekSame. And we close out with frontier tech, fusion, robotics, AI running locally on a laptop.

Speaker 3A lot of ground to cover.

DerekA lot of ground. Let's get into it, starting with the OpenAI filing. right now. Okay, so get this. OpenAI just filed to go public confidentially, and the Wall Street Journal broke the story this week, which I mean, nobody is shocked, but everybody is still kind of gaping at the screen.

Speaker 3Right, because we've all known this was coming eventually. The question was always when, and now the answer is, well, sort of now.

DerekSort of now. Because here's the thing. The WSJ piece actually quotes OpenAI saying, and I'm pulling this direct, it may be a while before they actually go public. They filed confidentially, which means no public S-1, no roadshow, no pricing.

Speaker 3So they filed, but they're also telling us not to get too excited.

DerekExactly, which is a very specific kind of energy to bring to a securities filing.

Speaker 3It's the we're going to the party, but we haven't decided if we're going inside move.

DerekYes. And look, there's a real strategic reason for that. OpenAI has said there are things they want to do that are quote, likely easier as a private company, the restructuring away from the nonprofit cap structure being the big one.

Speaker 3Wait, back up on that, because that restructuring is not a small thing. They're converting to a for-profit structure, and that has implications for the Microsoft relationship, the compute deals, all of it. Going public before that's clean would be... be messy.

DerekMessy is charitable. You'd basically be asking investors to price a company whose ownership structure is still being renegotiated.

Speaker 3And investors hate that, like the prospectus would have more asterisks than numbers.

DerekSo why file at all right now? That's what I keep coming back to.

Speaker 3I mean, there's a cynical read and a generous one. The generous read is they want to get the clock started on the SEC review process. It takes time. Filing early just makes logistical sense.

DerekAnd the cynical read?

Speaker 3The people who got in at lower valuations would like some liquidity at some point.

DerekYeah, that tracks.

Speaker 3Semafor had a really sharp piece on this. The concern flagged there is that these massive AI companies are going to need so much capital from public markets that they crowd out the fast followers, the smaller AI startups trying to raise or list behind them.

DerekThat's the part that doesn't get enough attention. It's not just OpenAI's IPO on its own. It's what the sheer scale of their capital needs does to the rest of the market.

Speaker 3Right. If OpenAI pulls in tens of billions in a public offering, that pool of investor dollars has to come from somewhere.

DerekAnd then you have companies on the other side actively pushing back on the pressure. Inc. ran a piece this week about a CEO of a data platform startup who straight up called this quick. This quote, "a terrible year to go public,"

Speaker 3He said that out loud?

DerekOn the record. His whole argument is the public market scrutiny doesn't match where his business actually is. And honestly, I've seen this exact conversation before. Founders get pressure from investors who want exits,

Speaker 3Ah!

Derekand some of them hold the line.

Speaker 3And some of them don't, and then the stock gets cut in half and everyone points fingers.

DerekYep, so you've got this really interesting split forming: the biggest name in AI signals it might eventually go public. go public, and some founders are looking at the market conditions and saying, hard pass.

Speaker 3Which raises the actual question everyone's dancing around. Are these AI valuations real? Like, can the numbers underneath actually support what investors are being asked to believe?

DerekThat's the test. OpenAI reportedly crossed a $100 billion valuation in private rounds.

Speaker 3Wow.

DerekGoing public means letting the whole market weigh in on that number, not just the investors who already said yes. Yes.

Speaker 3And that is a very different audience.

DerekMuch harder room to work.

Speaker 3You know what this reminds me of? Every major platform company that went public before its business model was fully baked. Some of them figured it out, some of them did not.

DerekThe accountability moment, that's what this IPO, whenever it actually happens, is going to force. Real numbers, real margins, real answers.

Speaker 3And speaking of companies being asked to finally deliver on their AI promises... There is one company with a billion plus devices in people's pockets that just had its own very public moment of reckoning this week.

DerekIs that a hint?

Speaker 3I mean, how many companies have that kind of installed base and still manage to make their AI assistant a punchline for two straight years?

DerekOh, we are absolutely going there. Shifting gears, let's talk about what happened at WWDC this week because Apple had a lot riding on Monday's keynote.

Speaker 3Yeah, and the framing from Reuters kind of says it all. The headline was literally Saving Siri. Two years of stumbles. So the question is, did they actually save it?

DerekOkay, so here's what they announced. According to TechCrunch's coverage by Morgan Little and Aisha Malik, Apple unveiled what they're calling Siri AI, a rebuilt Siri powered by Apple Intelligence with a pretty significant twist.

LaurenThe Gemini integration,

Derekright? The Gemini integration. Google's model running inside Apple's ecosystem, which is a sentence I never expected to type.

LaurenLike Apple, the company that built its entire identity around controlling every layer of the stack.

DerekAnd now they're saying, hey, for world knowledge and deeper queries, we're routing through Gemini, Apple's own press release called Siri AI profoundly more capable and personal.

LaurenI mean, they had to say something. But wait, what does the architecture actually look like? Because there's on-device processing and then there's the Gemini handoff. Those are pretty different things.

DerekRight, so the way MacRumors reported it, citing Juli Clover, iOS 27 is pairing on-device Apple Intelligence for personal context, things like your calendar, your messages, on-screen awareness, and then Gemini handles the heavier world knowledge queries, the idea being your private data stays local.

LaurenWhich is actually a smart split. For the non-engineers in the audience, think of it like this. Your phone handles the stuff that knows you, and Gemini handles the stuff that knows everything else.

DerekThat's a nice story, but show me the numbers, because Apple has made this kind of promise before, right? Two years ago, they demoed Siri features that just didn't ship.

LaurenYeah, yeah, that's fair. Reuters had a piece this week specifically on Apple's AI stumbles and whether this moment is different. Two years of missed timelines is a real pattern.

DerekSo what actually changes this time?

LaurenHonestly, I think the Gemini deal changes the calculus. Because before, Apple was trying to build all of this in-house, and they were clearly behind. Partnering with Google is basically admitting that.

DerekYeah, yeah, that's fair. Reuters had a piece this week specifically on Apple's AI stumbles and whether this moment is.

LaurenAnd it probably cost them. Like this isn't a charity arrangement. Apple is paying for Gemini access at scale.

DerekWhich, and I think this is the thread that runs through this whole episode. episode is another enormous check being written in the AI world. Apple integrating Gemini isn't just a product story, it's a business story.

LaurenExactly. And TechCrunch also noted iOS 27, macOS 27, the whole suite got updates built around this Siri AI push. It's the most significant Siri overhaul since the assistant launched.

DerekOkay, but here's my honest take. Apple's pitch is personal context plus world knowledge. knowledge. That sounds great. The question is whether it actually works reliably in daily use, or whether we're back to Siri misunderstanding half of what you say.

LaurenWhich has a proud tradition.

DerekA very proud tradition. Look, I've been on the wrong side of a Siri calendar appointment too many times.

LaurenI believe you.

DerekSo I'd say cautiously optimistic. The Gemini integration is a real change. On-device privacy framing is smart. But Apple has a trust deficit on Siri specifically, and that doesn't get fixed with one keynote.

LaurenAgreed. And honestly, the broader AI race is moving fast. Google has Gemini. OpenAI has its own assistant play. Microsoft has Copilot baked into everything. Apple needed to respond. Whether this is the response, we'll see when people actually use it.

DerekReal-world use is the test, always.

LaurenAlways. And speaking of enormous bets paying off, or at least being valued like they will, we've got a wave of funding rounds this week that are kind of staggering.

DerekOh, this is where it gets good, because Cyera just raised $600 million at a $12 billion valuation, up fourfold in 18 months.

LaurenFourfold in a year and a half?

DerekYeah, the AI security market is moving at a pace that should make everyone pay attention. Switching gears, the funding numbers this week are the kind that make you stop mid-sentence.

LaurenOkay, so Cyera, the data security startup, just raised $600 million at a $12 billion valuation. CTech reported that. And here's the part that actually made me do a double take.

DerekThe fourfold jump.

LaurenFourfold. In 18 months.

DerekWow.

LaurenThat's not a valuation bump. That's a different company.

DerekFrom an operator standpoint, when I see a number like that, my first question is always. Did the business actually grow fourfold or did the market just decide it was worth fourfold?

LaurenRight. And the honest answer is probably both, right? The security market is sprinting to keep up with AI adoption and Cyera's position right at that gap between what enterprises are deploying and what they can actually monitor.

DerekThat's a real problem that people will pay to solve, so I get it.

LaurenExactly.

DerekNow NinjaOne is interesting to me for a different reason. reason 400 million raised, 12.3 billion valuation Business Journal had that today infrastructure play

LaurenAnd this is the pattern, right? It's not just the flashy model companies getting the big checks. The picks and shovels layer is getting rewarded pretty aggressively right now.

Derekpicks and shovels classic

LaurenHey, it works. But seriously, if every company is becoming an AI company, then the tooling that manages those systems at scale, that's not a niche business. It's bet anymore.

DerekYou're essentially underwriting the whole AI build out when you fund infrastructure like that. The bet is infrastructure first, regardless of which models win.

LaurenSmart money, honestly.

DerekOkay, but the story I keep thinking about is just in earnest. TechCrunch had a piece on this today. He deployed nearly $500 million into startups like Anthropic, Anduril, SpaceX, and he did it without raising a traditional VC fund.

LaurenWait, back up. No fund? And how does that actually work?

DerekSo he built what TechCrunch described as a captive network of LPs. Instead of spending a year in fundraising mode, he lined up the capital relationships first and then deployed on a deal-by-deal basis much faster.

LaurenSo basically the fund structure itself is the inefficiency he cut out.

DerekThat's the argument. And when you're competing for allocations in rounds like Anthropic's,

LaurenYeah.

Derekspeed matters a lot.

LaurenIt does raise a question about accountability, though. Fund structures exist partly because they create alignment over a long horizon; deal-by-deal can get transactional.

DerekFair. I'd want to see what his LP relationships look like when a deal goes sideways.

LaurenThat's always where you find out who actually trusts each other.

DerekNow flip this on its head. All three of these stories point to the same thing. Capital is still moving fast, and it's moving into layers, security, infrastructure, and new fund structures.

LaurenAnd some of that capital is going somewhere that sounds like science fiction.

DerekOkay, so some of that capital Derek mentioned is landing somewhere that sounds like science fiction.

LaurenRight. So Helion just raised $465 million at a $15.5 billion valuation. This is This is a fusion power startup building an actual power plant for Microsoft.

DerekSam Altman is also a backer.

LaurenOf course he is. The man is everywhere. But here's what's interesting. TechCrunch reported they're targeting 2028 for delivery, like a working fusion power plant.

DerekOK, but what actually happened with the timeline? Because Fusion has been five years away for about 50 years.

LaurenFair. SiliconANGLE pegged the valuation at $15.5 billion, and I'm not dismissing it, but the question is whether they deliver electrons to a grid or just a very impressive press release by 2028.

DerekThat is the question every quarter from here on out. Show me the power. Switching gears slightly, Bloomberg reported Generalist AI hit a $2 billion valuation backed by Nvidia, $400 million raised led by Radical Ventures. This is robotics.

LaurenNVIDIA is backing a robotics startup called Generalist AI.

DerekThe naming committee really went for it.

LaurenI mean,

DerekShort pause.

Laurenshort pause. But NVIDIA's involvement matters. They're not writing checks randomly. If they're backing the robotics layer,

DerekYeah.

Laurenthey see it connecting to their chip stack.

DerekRight. It's not a bet on one company, it's a bet on the whole category needing NVIDIA hardware underneath it.

LaurenAnd then there's the Google story that I find genuinely fascinating. Gemma 4, their open source model, runs locally on a 16 gig laptop. VentureBeat covered it this week.

DerekWait, a full AI model on a laptop?

LaurenHandles audio, video, reasoning, no cloud required.

DerekWow.

LaurenAnd for enterprise, that's actually a big deal because data doesn't leave the building.

DerekThat's the part people underestimate. Compliance, latency, cost. Edge AI solves problems that cloud AI creates.

LaurenExactly. So zoom out for a second. We've got fusion reactors, walking robots, AI that runs on a laptop. Either we're at a genuinely extraordinary moment in the industry or the froth has gone fully vertical. Both things can be true. The real test next quarter is whether any of these bets deliver actual power to a grid or a robot on a factory floor.

DerekHard agree. That's where the story gets real. All right, that's a wrap on this one. And honestly, Derek, the moment that's still sitting with me is that Siri conversation.

LaurenSame. Two years of stumbles, one keynote, and we're supposed to just trust it now?

DerekApple has receipts to settle before we hand out trust, but the Gemini architecture split was genuinely interesting.

LaurenYeah, personal context local, world knowledge offloaded to Google. That framing actually made sense of it.

DerekAnd the IPO side, the big takeaway is just the hype and the timing are not aligned right now. OpenAI says it may be a while. Some founders are calling this a terrible year to list. The market's watching.

LaurenAnd that crowding out risk for smaller AI companies is real.

DerekOne hundred percent. Okay, if you got something out of today, subscribe wherever you listen, drop us a review and tag us on social if there's a founder we need to talk to. You need to talk to.

LaurenNew episodes every Wednesday. Thanks for being here.

DerekWe'll see you next week.