← All episodes
The Breakdown

New VC fund trying to break through, plus AI revenue numbers just went ballistic

Episode 2 · Thu 12 Nov 2026 · 16 min · The Breakdown

Transcript

Maya New VC fund trying to break through. That, and four more stories, on The Breakdown.

Intro

Alex This is Bits and Tickers. What got built, and what it's worth. I'm Alex.

Maya And I'm Maya. It's Thursday, November 12th, and this is The Breakdown: fewer stories, with time to dig into each one. Everything we talk about is linked in the show notes.

Alex On today's show. New VC fund trying to break through. AI revenue numbers just went ballistic. AI characters you can actually talk to. AI solves a really old quantum problem. And Nvidia's AI financing is getting serious attention. Let's get into it.

The Ticker

Maya Time for The Ticker. At Wednesday's close, the S&P 500 finished at 7,619, up 1.3 percent. The Nasdaq Composite was up 1.6 percent and the Dow was up 0.5 percent.

Alex And in big tech, Tesla had the best day, up 2.9 percent, while Microsoft was at the other end, down 2.3 percent. That's a snapshot, not advice.

Olaf Sakkers is betting he can beat the emerging-manager squeeze

Alex TechCrunch has a piece on this guy, Olaf Sakkers, who's trying to raise a second fund.

Maya Right. Apparently, it’s really hard to get in as a first-time fund manager these days.

Alex Seriously. PitchBook says established firms are grabbing almost everything – like 88% of the VC dollars this year.

Maya Since when? That leaves a tiny sliver for newcomers.

Alex It’s tough. He started RedBlue Capital in 2021, closed a first fund with about five million.

Maya Now he’s aiming for fifty million in the second. A big jump.

Alex Seems he’s been building trust, letting results speak for themselves, which he says takes time.

Maya He writes smaller checks, fifty to one hundred thousand, usually joining deals led by others.

Alex It’s interesting that he’s focusing on what he calls “the real economy,” like robotics and manufacturing.

Maya So, how does he actually get those LPs to commit, especially when the established firms are hoovering up all the cash?

Alex Apparently, it’s a trust thing. He says institutional investors are risk-averse, built to avoid zero returns, so they stick with the known brands.

Maya Go on.

Alex He had to prove it out with the first fund, backed by family offices, then gradually bring in institutions for the second. He’s deliberately letting the performance do the talking.

Maya That’s a long game. What was his background, anyway? Anthropology?

Alex Yeah, Princeton anthropology, a thesis on HIV in South Africa. Then he was drafted into the Israeli military – desk job liaising with the U.S. military, interestingly enough – and got into autonomous vehicles through that.

Maya Hold on, that's… a lot.

Alex It is. He was hired by Michael Granoff, who had capital to invest, and they built the first fund almost by accident, learning as they went. Granoff had been an early backer of Better Place, by the way.

Maya Huh. It sounds like he's leveraging connections.

Alex I mean, it's a smart strategy, letting the results do the talking. It’s a lot less stressful than chasing the big institutional commitments right away.

Maya It sounds like a lot of luck to me, though. Getting hired on the spot, building a fund "almost by accident"— that’s not repeatable.

Alex Maybe not, but he clearly has something – connections, a knack for spotting opportunities. Those five million dollars had to come from somewhere.

Maya I think it's more about inertia. Established funds have this gravitational pull, and it’s incredibly hard to break free from that.

Alex True, but he seems to understand the game. Small checks, letting others lead, providing support— it's a way to get his foot in the door.

Maya It’s still a long shot, though, isn't it? Getting to fifty million after starting with five?

Alex It's a test of endurance, he says.

Maya My read is that he's banking on the "gravity of perception" working in his favor eventually, but I suspect that’s the most fragile part of his plan.

OpenAI’s quarterly revenue reportedly crosses $10B for the first time

Maya Apparently, OpenAI crossed $10 billion in quarterly revenue. That’s according to TechCrunch.

Alex Hold on, what?

Maya Yeah, a nearly 50% jump from the second quarter. It’s a significant increase in their growth rate, too.

Alex Still, they weren't exactly struggling before.

Maya They were already growing at 18% previously. Their adjusted operating margin improved slightly, too, from -150% to -109%.

Alex That fast?

Maya They’re now talking about a $50 billion annual revenue run rate. Anthropic, meanwhile, is claiming an $81 billion run rate.

Alex Their calculation methods are different, though, right?

Maya Exactly. They count sales through cloud partners, while OpenAI doesn't for some of their sales.

Alex What’s driving this sudden acceleration, do you think? It sounds like a lot to pull out of nowhere.

Maya Apparently, they’ve been focused on tighter pricing, discounts, and better models. They're also building more enterprise-focused products and these AI agents.

Alex Really? I thought they were lagging behind Anthropic in that space.

Maya They were, but they’ve been trying to catch up, shifting focus to those areas. It’s interesting that they’re pulling ahead now.

Alex Just a slight improvement in operating margin though - still a massive loss.

Maya It narrowed from a 150% negative margin to 109%, yes. It's a small shift, but in that scale it’s still significant.

Alex $81 billion for Anthropic? Wow.

Maya That’s their annual revenue run rate, per their claim, and it’s due to those sales through cloud partners.

Maya I think it’s a bit misleading to get excited about this run rate stuff, honestly. It’s just a calculation.

Alex Hold on. Fifty billion? That’s still a lot of revenue.

Maya It is, but it’s still burning a ton of cash, Alex. That operating margin narrowing isn't exactly a victory.

Alex True, but they are shifting focus to enterprise, and that seems to be paying off, at least according to this.

Maya I’m less convinced. It sounds like they’re just trying to catch up to Anthropic now.

Alex I'm not so sure they're "catching up." They're just taking a different route.

Maya Maybe, but I still think the hype around these run rates is obscuring the fact that they're still losing money hand over fist.

Alex Ultimately, it just shows how much demand there is for what they're building, even if the current business model isn't sustainable.

Adobe’s AI experiments let you walk around in photos and interview your characters

Alex So, Adobe’s been showing off some experimental stuff at their Max conference – did you see anything that caught your eye?

Maya Project Screen Test, mostly. Apparently, you can now have conversations with AI-generated characters to, like, shape their personalities.

Alex Since when is that a thing? Sounds... involved.

Maya It's designed to give creators more control. You start by giving them personality traits, then chat with them to flesh out their backstory, goals, all that.

Alex And it generates animated tests based on those conversations?

Maya It does. My read is that it’s all about addressing the lack of creative control in AI video generation.

Alex Sounds like a bunch of demos.

Maya Still, it feels… clunky, doesn't it? I mean, what’s the actual process for giving the character a voice?

Alex You can upload sample audio, apparently. Or describe it in a prompt - they’ll try to match it. It’s not like they’re generating from scratch.

Maya So you're basically building a digital puppet?

Alex Kind of. The article mentioned it’s because existing AI video tools lack creative control; traditional animation has absolute intention at every step.

Maya Good to know. Sounds like a neat trick, though.

Alex Definitely a demo. I'm not seeing how that becomes a workflow.

Alex It's a clever way to inject some personality, I suppose, but it still feels like a lot of steps for something that could be done more simply.

Maya It's not about being simple; it's about giving creators a tool that wasn't there before.

Alex But is it a useful tool? It sounds like a hobby.

Maya I think it addresses a real pain point in the video generation space. The article explicitly said so.

Alex Maybe. Still, building a digital puppet… it’s a weird framing.

Maya It’s a metaphor, Alex. It's illustrating the level of control.

Alex A metaphor that doesn't really translate to a practical workflow, in my view.

Maya It's a first iteration, and even in that state, it's demonstrably an improvement over what’s available now, so it’s already valuable to some.

PSI is trying to make a business out of AI-driven scientific discovery

Maya According to TechCrunch, a startup called PSI used an Anthropic model to crack an 18-year-old problem in quantum computing.

Alex Hang on. Seriously?

Maya Yeah, a recent MIT grad, David Turturean, apparently did it. He’s deferred his master’s to work at PSI.

Alex Wow.

Maya The problem has to do with how quickly quantum computers can solve certain issues, and Scott Aaronson, one of the original researchers, says he’s… well, he's got a lot of feelings about it.

Alex Is that all?

Maya It’s a negative result, apparently closing one door on quantum computing, and PSI’s CEO, Matt Pines, is saying theoretical physics is going to be mostly machines’ work soon.

Alex Bold claim.

Maya Their current revenue comes from optimizing data centers – simulating things like heat and power – but the real goal is to turn these breakthroughs into actual tech.

Alex How did he even do it? Just… a recent grad and an Anthropic model solving something that’s stumped people for 18 years?

Maya Well, it wasn't just any Anthropic model. It was Fable 5.2, an unreleased one. And he used what they’re calling specialized model harnesses and scaffolding, which apparently is key.

Alex Harnesses? What are those?

Maya It’s like… tools and memory systems, helping the model really dig into the problem. The CEO, Pines, says they need a lot of that to tackle the work properly.

Alex Feels like a demo, doesn’t it?

Maya Aaronson, one of the original researchers, seemed to think so too, initially. He was surprised it hadn’t shown up in OpenAI’s Mathocalypse, wondering if it was so much harder.

Alex So, is PSI just doing this for fun?

Maya Not exactly. They’re currently making money optimizing data centers, but Pines is pitching the longer-term vision: turning these theoretical physics breakthroughs into actual technology. He says he had to stretch to convince investors six to nine months ago that this was possible.

Maya I think Pines is just playing the investor game, really. Saying theoretical physics will be mostly machines’ work soon? That’s a sales pitch.

Alex Hmm. It’s impressive, though, isn’t it? A recent grad solving something that’s been open for so long.

Maya Sure, impressive. But it's a demo. Aaronson himself seemed to think that.

Alex But the harnesses and scaffolding thing… that does suggest something more than just throwing a model at a problem.

Maya Maybe, but ultimately, it’s still just one result. They're chasing a long game, and data center optimization is the real money-maker right now.

Alex That's true, but the fact that they were able to crack something so fundamental, even with a specialized setup… it indicates a potential shift.

Maya I’m not convinced it’s a shift; it's a carefully managed narrative.

Alex Still, someone built something that could do this. And that’s a worthwhile accomplishment in itself.

Analysis: Is Nvidia too big to fail? Its AI financing web raises the stakes

Alex This CNBC piece about Nvidia being, essentially, the central bank of AI, caught my eye because it’s a fascinating way to think about how interconnected everything has become. Apparently, they’ve made huge financial commitments to other firms to help grow the industry.

Maya Hundreds of billions? That’s…

Alex Yeah, and it’s got the Federal Reserve and other financial institutions watching closely, just in case something unexpected happens. They’re worried about a "broad-based repricing of assets," apparently.

Maya Hold on, so is someone saying we’re headed for another 2008 situation?

Alex Sheila Bair, a former regulator, says Nvidia isn’t "too big to fail" in the same way a bank is. They still make a product people want, and have a big equity buffer.

Maya Makes sense. But she does say regulators should still watch for exposure across the financial system.

Alex And then there's the whole FBI and Congress thing, reportedly concerning Huang's relationship with someone. The board still has confidence in him, though.

Maya It’s a lot to juggle, even for Nvidia.

Maya So, how exactly does this "central bank of AI" thing work? Just… handing out money?

Alex Pretty much. They’re providing financial commitments, essentially loans or guarantees, to other companies to help them build out their AI infrastructure. It's not just cash, it's a whole web of financial support.

Maya Hold on, what happens if those companies can’t pay back those commitments?

Alex That’s the worry. If Nvidia’s GPU chip values drop, it could trigger a chain reaction where companies struggle to repay what they borrowed to buy them.

Maya That’s… concerning.

Alex Sheila Bair pointed out that even though Nvidia isn't structurally the same as a bank, that ripple effect is what regulators are paying attention to – the exposure across the wider system.

Maya And the Huang situation complicates things, doesn’t it? It’s a weird timing.

Alex It certainly does. The article mentions analysts are now considering a range of outcomes, including his potential removal, which would be… quite a shift.

Alex I think the “central bank of AI” analogy is really underselling the risk here. It’s not about Nvidia failing outright, it’s about those financial commitments suddenly becoming a problem for a lot of other companies.

Maya I see it more as a symptom than a cause. Nvidia is just capitalizing on the existing demand, providing the financing that already exists somewhere else in the system.

Alex Still, a lot of money is tied up in their continued success. Huang’s situation adds a layer of unpredictable volatility, doesn't it?

Maya It's certainly distracting. But the board's confidence in him is important to note, it’s not like they’re in crisis mode.

Alex True, but that doesn't negate the potential for a broad-based repricing. It’s a delicate balance.

Maya My read is that the real concern isn’t Nvidia itself collapsing, but the interconnectedness of the whole AI buildout finally revealing its fragility.

Alex Makes sense.

Maya The worry isn't that Nvidia will fail, but that its success has masked a systemic reliance on a single point of potential failure.

The Takeaway

Alex That's the news. Time for The Takeaway.

Maya I think the common thread here is just how much money is sloshing around, enabling all this experimentation, right?

Alex I'll be watching to see how quickly those Adobe AI features actually ship – demos are easy.

Maya I'm curious to see if PSI can actually translate their AI discovery efforts into something beyond a research grant.

Alex Hold on, I'm not writing off Olaf Sakkers' approach just because it's niche. Someone’s gotta find the inefficiencies.

Maya I still maintain Nvidia’s financing web isn’t that surprising, Alex. It's just… infrastructure.

Alex I guess we’ll agree to disagree on how much that signifies about systemic risk.

Maya And that's The Breakdown. Every story we talked about is linked in the show notes.

Alex Next up is The Long Drive, tomorrow. One reminder: nothing here is financial advice. From Bits and Tickers, I'm Alex. Keep building.

Maya I'm Maya. And keep counting.

Show notes

In this episode:

Market figures: Yahoo Finance closing prices.

Got a question for the show? Send it in and we may answer it on First Tick.

Nothing in this episode is financial advice.