Invested by Aleph · 2026-08-12 · 16 min
Key moments - from our scoring
Substance score
58 / 100
Five dimensions, 20 points each
This special Invested episode assembles conflicting expert perspectives on artificial intelligence recorded across multiple epochs of rapid AI development. The guests - including Gavin Baker, Miha Kaufman from Fiverr, Ellad Raz, Aaron Sheeran from Nexar (merged with Nauto), David Magerman, and others - debate whether current GPU spending represents a Ponzi scheme or a genuinely profitable infrastructure play. Core disagreements center on whether compute costs are actually improving (Baker argues Nvidia's Hopper to Blackwell shows only linear 2x scaling in power consumption), whether large language models will remain the dominant approach, and whether AI will increase or decrease employment. The episode explores the difference between training and inference economics, the debate over open-source versus proprietary models, and whether a factor-of-10-to-50 financial bubble exists despite real economic value creation. Speakers discuss practical implementation: using AI agents to scale operations, bonusing developers by token consumption on Claude and Gemini, and whether AI will create mass unemployment or productivity gains across industries.
Experts disagree sharply: some argue it's a factor-of-10-to-50 larger financial bubble than the dot-com era with monopoly-like dynamics (10 companies funding each other's purchases), while others point to genuinely positive ROI on $200B, $600B, and early trillion-dollar spending, though margins are tightening as capital rises.
No - Blackwell shows only 2x performance improvement paired with 2x higher power consumption compared to Hopper, representing linear rather than exponential scaling with no architectural innovations between generations.
Current data shows companies investing heavily in AI are actually hiring more people and becoming more productive, contradicting earlier predictions of mass job displacement, though whether this represents true job creation or interim dislocation remains unknown.
The answer is unclear; while frontier labs (OpenAI, Anthropic, Google) control training, their business models lack sustainable unit economics, while open-source models now compete with frontier outputs, and companies using AI for inference are seeing profitable margins.
Avoid betting on LLM capability freezes or assuming models won't improve; instead, identify specific data sets with decision-relevant information and match them to appropriately complex tools rather than defaulting all problems to large language models.
Our reviewer’s read on each dimension, with quotes from the episode.
The episode contains a mix of substantive observations (ROI calculations at scale, the distinction between training and inference profitability, token-based developer incentives) alongside significant padding and meta-commentary. Speaker F spends considerable time on framing and housekeeping rather than substantive content, diluting the insights-per-minute ratio. The guest remarks are punchy but often lack depth.
All of the biggest spenders on GPUs, with the exception of, um, XAI, are public companies. And public companies report something called quarterly financials...the answer to the $600 billion question was yes, the ROI has been awesome. Now we're at the trillion dollar question and the ROI has actually still been good, but we're rapidly heading for 2 to 3 trillion.
The driving force here is data. And it's not just data, it's information. And so just because you have data doesn't mean you have a solution.
The episode recycles familiar AI debates (bubble vs. genuine transformation, job creation vs. displacement, frontier labs vs. open-source competition) without offering fresh frameworks or first-principles analysis. Most claims reflect consensus positions in tech media circa 2024. The observation about exponential misunderstanding is well-worn, and the compute hierarchy argument lacks novelty.
So much is changing in this world every day that it's literally hard to compute.
We're really bad at understanding exponents. And this is this kind of inflection point where things will get compounded and get, you know, crazier and crazier from here on out.
The episode features credible practitioners (Gavin Baker, a fund manager; someone from Fiverr; a chip architect discussing Hopper/Blackwell; someone running developer AI incentives at scale). However, it's a compilation format with short snippets, making it impossible to assess depth of expertise. Most speakers are identified only by their role or company, not by their track record. The absence of names and biographical context limits caliber assessment.
You'll hear on this compilation from the great Gavin Baker, who has certainly been bullish.
You hear from Aaron Sheeran, Zach Greenberger of Nexar that has since merged with Nauto to create the greatest data lake outside of Tesla and model of the physical world
The episode includes concrete numbers (factor of 10-50 for bubble scale, $10B vs. $15 cost structures, $200B/$600B/$1T spending thresholds, $300-$100 per developer token costs, Hopper/Blackwell 2x scaling). However, many claims lack specificity: no named companies beyond Nvidia, OpenAI, Tesla, Fiverr; no timestamps or sources for quarterly financials cited; vague references to 'a couple of million people' using AI at the bleeding edge. Data points exist but feel cherry-picked without systematic evidence.
When you buy, uh, $10 billion worth of computers, um, from Nvidia and you invest in Nvidia, you're basically just paying yourself for equipment
Now we're at the trillion dollar question and the ROI has actually still been good, but we're rapidly heading for 2 to 3 trillion.
Speaker F (the host) asks few probing follow-ups and mostly acts as a traffic director, stitching together guest clips with heavy framing. When questions do appear, they're often softball or rhetorical ("What do you think that means...philosophically?"). There's minimal genuine push-back or disagreement aired - mostly agreement that things are moving fast. The format itself (compilation of brief clips) prevents sustained conversational depth or challenging exchanges.
F: If you're interested in finding out ahead of time what's going on in Israeli tech...then you want to subscribe to Aleph's newsletter
F: You're bonusing people to rack up token bills on Cloako? E: Yes, Correct. Crazy, right?
Computed from the transcript - who did the talking, and the words that came up most.
On this special episode of Invested, Michael Eisenberg pulls together the sharpest AI arguments from the last year of the show and puts them in conversation with one another. You'll hear from Gavin Baker, Sarah Tavel, Micha Kaufman, Elad Raz, Eran Shir, Zach Greenberger and David Magerman - seven investors and founders who disagree about almost everything except the stakes. This is not a group of people arguing over the same set of facts. Because the field is moving so quickly, each of them is describing a different epoch of AI. Compute was supposed to depreciate quickly - instead, demand is growing. Inference wasn't supposed to be profitable - today it is wildly profitable. AI was supposed to be a game for frontier labs only - open source models are now competing with them. And AI was supposed to trigger mass unemployment - the latest numbers suggest the companies going all in on AI are hiring more people, not fewer.
Transcribed and scored by The B2B Podcast Index.
Speaker A: AI is an automatic machine gun, right? And we're giving it to children.
Speaker B: I think we will have artificial superintelligence in 10 years. And by that, I mean an AI that is smarter than the smartest human
Speaker C: in every discipline building AI accelerator just to run transformers. Dumbest ideas ever.
Speaker D: If your startup assumes that LLMs are going to stand still, we're going to bulldoze you.
Speaker E: The only way to deal with AI is AI. What was considered to be hard is going to be the new simple. And what was considered to be almost impossible is going to be the new hard.
Speaker F: Over the last year, we've done a number of episodes that have touched on the topic of AI artificial intelligence. You might think when you watch this compilation that these are the same facts, but things are moving so quickly in this era of AI that the different people speaking in this video are actually talking about different epochs in the great debate on AI. You'll hear on this compilation from the great Gavin Baker, who has certainly been bullish. Not only has he been bullish, when the market went down in July, he flew out to Silicon Valley to see whether he was missing something. And he thinks he's not. In fact, he thinks he's underestimated just how powerful this AI trend is. You'll hear from Sarah Table I interviewed a while ago, Miha Kaufman at Fiverr, whose stock has been decimated by the AI narrative out there. You hear from Ellad Raz claiming he's gonna take on Nvidia. You'll hear from Aaron Sheeran, Zach Greenberger of Nexar that has since merged with Nauto to create the greatest data lake outside of Tesla and model of the physical world outside of what Tesla, uh, is building. And you'll also hear from David Megerman, who I would say said AI has been here a long time, and all this is a giant bubble. Well, we might be in a bubble, and we might also be in the largest transformation of our lives of the last hundred years. And we're only going to know in 10 years where we are in this long process of artificial intelligence. There's so much out there. This is such a rich topic that's moving so fast. We need to remind ourselves how many of our assumptions not long ago were incorrect. There have been a small number of great investors that have figured out where the puck is going, as the great hockey player Wayne Gretzky once said. And so I really want you to listen to this, to realize how far we've come in what's going on in artificial intelligence, how so many People were wrong that compute would depreciate quickly. In fact, it's growing. How wrong so many people were that inference couldn't be profitable. It's wildly profitable right now. How little we understood a year ago about the difference between training and inference. How much we thought that this was going to be a game for only Frontier Labs. And here we are with open source models competing with the Frontier Labs. So much is changing in this world every day that it's literally hard to compute.
Speaker A: The scale of the bubble, the financial bubble. I would say that the bubble that exists today is probably a factor of 10 to 50 of what the Internet bubble was. But the problem is it's monopoly money. It's these like 10 companies giving each other billions of dollars. When you buy, uh, $10 billion worth of computers, um, from Nvidia and you invest in Nvidia, you're basically just paying yourself for equipment and so you can overpay yourself and, and it all kind of cancels out if it's costing the company that's doing the inferencing $10, eventually they have to charge 15. OpenAI isn't a functioning business. Yeah, it's not. I mean, the business model is so demonstrably, uh, a Ponzi scheme that, um, I don't know when it ends.
Speaker C: Take the core compute of Blackwell, core compute of Hopper. You compare them one by one. You see Blackwell is 2x than Hopper.
Speaker A: Mhm.
Speaker C: Wait, it also consumed 2x more power.
Speaker E: Mhm.
Speaker C: Nothing in the architecture had been improved between Hopper and Blackwell.
Speaker F: Mhm.
Speaker C: And willing for sure. Other nuances, but that's like their core. Hopper to Rubin.
Speaker E: Mhm.
Speaker C: 2x more power. Mhm 2x more performance.
Speaker F: Mhm. This is linear scaling.
Speaker C: Linear scaling.
Speaker B: All of the biggest spenders on GPUs, with the exception of, um, XAI, are public companies. And public companies report something called quarterly financials. And you can use those documents to calculate something called return on invested capital. Um, it's really easy to do. And to date, the ROI on all of this spending has been strongly positive to be good. So while the answer to the $200 billion question was yes, the ROI was awesome, the answer to the $600 billion question was yes, the ROI has been awesome. Now we're at the trillion dollar question and the ROI has actually still been good, but we're rapidly heading for 2 to 3 trillion. And will the ROI continue to be good? Unknown. And if it's not, do economics dictate or does the prisoner's dilemma dictate and the companies could keep spending.
Speaker D: I started at Bessemer in 2006. I've never seen anything like this. You know, SaaS was incredible, mobile was incredible. But this just feels very different. If the technology froze today, there's still so much impact that we're going to see that it's just incredibly exciting. Of course, there's the variable inference costs and that for many of the labs, is gross margin profitable. But then there's the amortizing, the cost of the model training, and that's where the numbers don't look as pretty for these, for these firms, but should, over time, be okay. At least that's the belief that was
Speaker F: the argument about whether we're in a bubble. And by the way, you could be in a bubble and still there could be incredible economic value created by the small number of companies that survived the bubble. You've, of course, read and heard about the great debate on employment. Everybody thought AI was going to create mass unemployment. Well, the latest numbers show that actually AI is increasing employment. Those companies that are all in on AI are actually hiring more people and they're becoming more productive. So what is it? Will it create mass unemployment? Is it just a shift in employment with, let's say, some interim dislocation? Or is it going to create more employment like every other technological revolution in the past? The answer is we don't know. What's important when being an investor and listening to a podcast called Invested is whether the people bet their convictions, whether you're willing to bet your convictions, whether willing to invest confidently in what you believe about the future of this transformational technology that will touch every part of the economy and every country on planet Earth. And the most interesting thing is, it's the early innings. It's literally the first inning of, of AI. There's maybe, maybe a couple of million people around the world using AI at the bleeding edge. Maybe a couple of million people. There's 8 billion people in the world. There is a long way for this revolution to run.
Speaker E: AI is creating an illusion for most people because it makes us better, it makes us more productive. And you think that you're gaining these superpowers as a human being. It makes you more confident and you can do more stuff. Yes, but it's giving the same power to everyone else, which means that it doesn't give any power to anyone. I'll help anyone who helps themselves, but if you look for me to educate you about AI, uh, you're doomed. You're done. You're done. And your problem is you're not Done at fiverr, you're done. Period. There will be no demand for lazy people that expect the world or think that the world owes them anything.
Speaker D: The economic value proposition, we haven't seen anything like it. Maybe since virtualization, all this software as a service that has always been this kind of squishier. I'm going to improve the productivity of your existing employees and how much more? 10%, 20%. Now with AI, it is not even close how compelling the value proposition is, how clear it is.
Speaker B: But in a world of AI, fundamentally the economic return, a lot of it will come from either replacing or augmenting humans with GPUs. Labor replacement or augmentation. The maximalists I think for sure would say labor replacement. So I think that's a fair way to calculate it.
Speaker A: The best we can hope for, I think from these tools is replacing junior level work. And the way that you get senior people, the promotion, the uh, mentorship, um, the ability to create a mature workforce is gonna be severely stunted by, uh, the use of AI.
Speaker E: You know, everyone talks about resources in terms of people, but resources are now being talked about in terms of agents. So you don't necessarily need to scale your human capital, you can scale your agent capital. So we have agents running across the entire company that are doing everything from product design to product deployment to creative.
Speaker F: You're bonusing people to rack up token bills on Cloako?
Speaker E: Yes, Correct. Crazy, right?
Speaker F: So the incentive is to create as much code as possible, whether it works or not.
Speaker E: Well, phase one. That's phase one.
Speaker C: That's phase one dot literally see people getting 10x more productivity using those tools.
Speaker F: Have you mandated that everyone needs to use.
Speaker C: Absolutely.
Speaker F: AI coding.
Speaker C: Of course. I mean a company that don't have the leaderboard scores of how many tokens is being used is the more the merrier. Of course.
Speaker F: How big is your token budget now?
Speaker C: Depends. Like we are using everyone. Me personally, I'm using Gemini, Codex and Claude. Uh, we have people that generate $300 a day, $100 a day per developer.
Speaker E: Ideally, I want everyone to automate 100% of what you're doing. And people looked at me and said, um, so why would you need me? And I said, great, because now you have 100% of your time free to do interesting stuff, things that machines cannot do. Yet.
Speaker F: If you're interested in finding out ahead of time what's going on in Israeli tech, what the latest trends are, who's breaking not only news, but breaking the trend lines, then you want to subscribe to Aleph's newsletter and you can sign up right in the show notes in the description of the show. Please subscribe to the newsletter. One of the disagreements or arguments that people lose sleepover is who will AI pay off for? Not if it will pay off, but for who, because there's a lot of dollars at stake. The question I'm sure you're all asking yourself, and I ask myself is what's going to happen next? But you think the world's gonna change a lot in the next nine years.
Speaker E: Yeah.
Speaker F: Even though over the last nine years it looked the same.
Speaker E: Yeah. Because we're really bad at understanding exponents.
Speaker C: Right.
Speaker E: And this is this kind of inflection point where things will get compounded and get, you know, crazier and crazier from here on out.
Speaker D: Sam Altman in a, uh, podcast, I think it was with Harry Stevinks, so said. If your startup assumes that LLMs are going to stand still right now, we're going to bulldoze you. And I think the biggest mistake that I see people make is either they're building something that isn't a beneficiary of, uh, the kind of continued and astounding march forward with the power of these, of these large language models. And also if you are building something and then it's like a bridge, uh, because you're assuming that the LLMs or whatever the labs are doing isn't going to progress forward and you're going to outrun them, uh, that feels like a very, uh, treacherous, uh, game to play.
Speaker E: It's cheaper to do it by hand. It's still going to be manual work and if it's not, it's going to be replaced by technology. The types of repetitive stuff that even machines cannot do and still require human are just going to change. They're not going to go away.
Speaker A: The driving force here is data.
Speaker F: Mhm.
Speaker A: And it's not just data, it's information. And so just because you have data doesn't mean you have a solution. And, and if you have data that has information in it, from an information theory perspective, there are dozens of mathematical algorithms and pattern recognition algorithms that can pull that information out and help you reduce uncertainty about the future. So we've become a one trick pony where everything is just thrown into an LLM. I think success is looking at the data sets that you have, figuring out how much information you have and which information that you have is relevant to solving a real world problem and then finding the right tool at the right level of complexity and cost to extract that information.
Speaker F: What do you think that means not just financially, but philosophically about the world where this notion of silicon, um, could become the most valuable thing in human civilization?
Speaker C: No surprise there. I mean, think about it for a second. When 1,000 years ago, we all were fighting around food. That's the basic element that everyone needs. Food, water, resources. I think that around 300, 200 years ago, it was around energy resources.
Speaker E: Mhm.
Speaker C: That's most of the world. Now we are in the age of compute.
Speaker B: But I think one of the most interesting parts of all of this is we don't know what the economic returns to superintelligence will be, definitionally, because we have never seen it. And if humans, we have pushed the limits with, you know, people like Albert Einstein or whoever else, push the limits of the natural laws of the universe, if we push the limits of physics and biology and chemistry and the math that underpins all of it, then the economic returns to superintelligence, they may not be that high. On the other hand, if we haven't and superintelligences are curing cancer, inventing warp drives, and we're colonizing multiple solar systems, the returns to superintelligence are going to be really high. But it's very unknowable.
Speaker F: Thank you for listening to this special episode, Invested. I hope you're left with more questions than answers. I hope you'll continue to deep dive and really try to understand to the extent we can, where AI is going, what's it going to impact? How do you invest in it successfully? And I hope this will be a smorgasbord of opinions that you can choose from and at the same time, just an appetizer and you'll choose to go deeper. Thank you so much for tuning in to this special summer edition of Invested. If you enjoyed the podcast, please rate us five stars on Spotify and on Apple Podcasts or wherever it is that you listen to our podcast. Please subscribe to our YouTube channel and please, please, please subscribe to our newsletter. That way you'll get the episodes for everybody else.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.
WTT: AI: Fundamentals, Valuation, and the Next Allocator DilemmaCapital Allocators · features Gavin Baker56 / 100
Ep 160: Box CEO Aaron Levie on Silicon Valley's AI Problem, Evidence for AI Job Creation & What Doomers and Accelerationists Get WrongJoe Lonsdale: American Optimist · on Open source AI models88 / 100
AI Is Producing More Code, but Is It Producing More Value?The Data Exchange with Ben Lorica · on Open source AI models88 / 100
Less about Models; More about ArchitecturePractical AI · on Large Language Models (LLMs)85 / 100
Microsoft Fabric: The Platform That Turns Data into Competitive AdvantageLeading IT - APAC Insights · on Large Language Models (LLMs)85 / 100
Episode 7: AI & the Power of a "Thin Core"Architecting the AI Enterprise · on Large Language Models (LLMs)82 / 100
Yuval Ariav, Solo GP of Symbol Capital, on Pre-Consensus Investing, Why NextSilicon Could Be Israel's NVIDIA, and the $1.1 Trillion Market Hiding in Plain Sight86 / 100
Daniel Lubetzky, Founder of KIND, on Reaching a $7 Billion Valuation, Why Builders Beat Destroyers, and How Many Nickels It Costs to Stick to Values
Saar Safra, Founder & CEO of Beewise, on Why Bees are a Matter of National Security, the Honey Laundering Scandal, and the AI Beekeeper That Never Sleeps
David Siegel, Former CEO of Meetup & Investopedia, on Solving the Loneliness Crisis, Rescuing Meetup from WeWork, and Why Israelis are Happier than Americans
Oren Zeev, Solo VC Managing $3B, on Why Code Is No Longer a Moat, Surviving Navan's Worst IPO of 2025, and the Dangerous 10x Growth Mindset