77 – The Great Talent Redistribution
The Great Talent Redistribution: Where is Talent Actually Going in 2026 and beyond? Is the start-up compensation model broken? How about big Big Tech? How about non-tech small & medium businesses? What is happening to talent, going forward? This and many other topics in this episode of Tech Deciphered. Navigation: Intro The Broken Contract? The Great Unbundling The Three (?) Destinations Alternative Cap Tables, Alternative Compensation Models Investor Landscape Fragmentation Operator Playbook and Predictions Conclusion Our co-hosts: Bertrand Schmitt, Entrepreneur in Residence at Red River West, co-founder of App Annie / Data.ai, business angel, advisor to startups and VC funds, @bschmitt Nuno Goncalves Pedro, Investor, Managing Partner, Founder at Chamaeleon, @ngpedro Our show: Tech DECIPHERED brings you the Entrepreneur and Investor views on Big Tech, VC and Start-up news, opinion pieces and research. We decipher their meaning, and add inside knowledge and context. Being nerds, we also discuss the latest gadgets and pop culture news Subscribe To Our Podcast Nuno Goncalves Pedro Introduction Welcome to episode 77 of Tech Deciphered. This episode will focus on the great talent redistribution. Where’s talent actually going in 2026 and beyond? The Silicon Valley deal of the last 30 years, very low salary, stock options, you will either sell for a ton of money or IPO, and everyone gets rich, is seemingly broken. Or is it really? The dominant narrative says the tech middle class is dying. We disagree. There is obviously a lot of stuff going on whereby big tech is partially barbelling. There’s a superstar concentration on the top. There’s a bit of a seemingly allowing of the belly. We’ll come back to that. We don’t quite believe that is totally true. There’s a collapse at entry level. The belly is migrating into three, potentially even more, very different destinations: AI native startups, human-verified premium businesses, and the read the industrialized middle of the S&P 500 and SMB world. Each has its own cap table, each will have its own compensation model, and each will have its own investor profile. In some ways, this is the third episode in our Reset trilogy. We started with episode 75 on the SaaS-apocalypse. We talked about the great private capital reset in episode 76, and now we talk about talent redistributions. Bertrand, exciting times, not always positive times. Bertrand Schmitt Yeah, it’s exciting times because it’s a time of change. Of course, we have the doomsayers. If you listen to Dario Amodei of Anthropic, every white-collar job on Earth is going to disappear. I think I strongly disagree, and I suppose you too as well, we strongly disagree. It’s going to be more of a redistribution. If you look at the history of technology, this is what always happened. We forget how many jobs have disappeared over the past 150 years. We move from a time of 150 years ago. People were mostly in agriculture. Then you had a lot of weird jobs that disappeared from people transporting water to people bringing ice from the pools to people doing the job of computers. People forget that computer was a title given to human beings. We’re doing calculations. Then, of course, secretory jobs in the ’80s, ’90s, where suddenly anyone can type using a word processor, the rise of Excel, that sort of stuff. Many things have changed. Some jobs have indeed disappeared. Some jobs have totally transformed. Where you do these jobs have changed. I think we are at a similar stage where, thanks to AI, and I would say for now, or at least the rise of AI coding, there is a dramatic change happening. I don’t think it means that people will be without a job. It just means, from my perspective, that jobs are changing. You are not just doing a lowly coding level task that actually indeed could be replaced, but you are going to have more of builder type of mindset, a product manager type of mindset going forward. We also expect that the distribution of jobs, depending on the type of business, will be quite different. Nuno Goncalves Pedro The Broken Contract? Maybe let’s reset a little bit to the broken contract, or if it’s really a broken contract. There’s been this image in technology and tech that basically you get paid very little to work in tech. You get a bunch of stock options. The earlier you are in the company, the higher the level of stock option grants you get. Then you make a ton of money at some point because the company will either sell or IPO, and that’s heard of it. Obviously, there’s a lot of movements happening right now that are changing how these dynamics work. The first part is obviously AI, and in some ways, AI is shrinking companies. It’s not unheard of that companies with as little as four or five people reach 50 million in ARR. There’s companies with one person that have gotten bought for hundreds of millions of dollars or billion of dollars. Obviously, things are moving very, very fast, and therefore, there isn’t a large employee cap table. How would you share the upside? Would you actually give a couple of percentage points to an early employee rather than your 0.2-0.5% kind of thing for early employees? The second part is a little bit the other side of the table, which is the IPO market is seemingly in a drought. There’s not much happening in IPOs. Maybe 2026, at some point, there will be an unlock, but right now, it’s seemingly difficult to get your upside. Even if you’re an employee, you have to wait a long time. The median time of IPO has climbed over 10, 11 years, the longest in over a decade. Basically, not only you have to wait a long time as if there is an IPO drought, like we might be going through right now, when do I actually get my cash back? Unless the company gets bought, maybe there are secondary transactions along the way, maybe there’s something else. But obviously there’s a little bit of a reduction and lowering of the upside seemingly for this contract and for this place. The easy conclusion that I think many are taking is, because of all of this and all the layoffs that are happening, even in big tech, that serve the tech middle class is dying, that basically AI screwing the workers, et cetera, there’s also a lot of discussion that even it might be affecting the entry-level jobs as well. Everyone coming out of undergrad right now can’t get a job, et cetera. There’s this doomsday scenario that you’re alluding to that everything is changing. We have a slightly different perspective. We think there’s a realignment of market. In layoffs, there was a lot of layoffs that were warranted. Big tech, in particular, had actually hoarded a lot of engineering capacity over the last decade or so. There’s a little bit of a realignment that needed to happen in any case. When everyone’s saying, “Well, AI is compressing everything,” well, it’s compressing right now, but we don’t think actually it’s going to compress over time. You’ll still need engineering and science talent to come on board for you to be able to scale up. It’s not like AI is going to take care of everything and teams are going to be five people for companies that are worth a trillion dollars. That’s not happening. Today’s thesis, I think a little bit of this doomsday scenario needs to be seen with a more nuanced lens. I think that’s how we’re framing today’s episode, that there’s a bit of a nuance, there are some extremes happening. We’re going to talk about those extremes, but ultimately, it’s not quite as simple as saying that the tech middle class is disappearing in early jobs are going to be a thing of the past. Bertrand Schmitt At the same time, what you started with is true. I mean, that 50 million ARR company, just five people. At a bigger scale, that’s exactly the matrix for Anthropic. They have reached a stage where they are at a range of 12 million ARR per staff per employee. It’s metrics that are definitely never seen before. I don’t think any company raised to this level. Best in class, best run companies, one, two million per employees. I mean, that was your target if you can make it. We are definitely in a different game. But I think what matters at the end of the day, and that’s what we’re arguing, is that you have to see the big pictures. Yes, some positions might disappear inside some companies, but some other positions will be created in other companies. Usually, what people do is keep talking about the jobs who disappear and not looking at the bigger picture of jobs that are being created as well. What is true, and I think you alluded to that, is that the big tech the past 10, 15 years had some strategy of hoarding talent in a war where having the best talented people will make the difference in numbers, will make the difference between winning or losing. The Google of the world, the Microsoft of the world, the Amazon of the world, they were hoarding talent. They would try to make sure that they might not have such needs in talented number of people. But if they have the talent, it means their competitors didn’t have the talent. It means that the startup trying to reach scale couldn’t pay the giant salaries that the Google of the world were paying. There was definitely some hoarding. But it went so far in the 2020, 2021, that I think since then there has been a coming back to normal. There is also now in 2026, the recognition that it’s not true anymore. Yes, talent can be very valuable, but there is now a bigger and bigger gap between the extremely talented versus the rest that are merely talented because of AI. AI is able to replace at scale your software engineers, your software managers. I would say it’s quite new. I don’t think it was true a year ago. We’re really talking about a recent dramatic change in what can be achieved thanks to AI. We can see most of the big AI companies are moving to coding. It was started by Anthropic as a trend, OpenAI has followed through. Obviously, the Cursor of the world existed before, but they were not as successful. All the Chinese open-source models are moving very fast to coding optimization the past few weeks. It’s quite an incredible change. I think there is that dramatic change, recognition that coding can be done differently. As a result, we are going to see change in the distribution of jobs. I think it will start from the top because we see the news of the big Google, Microsoft, Amazon, and others who used to hold talented software developers to a change in realization that no, we actually need to invest in AI. We need to invest in compute because compute is going to do the job of most of these people. Therefore, we can’t pay for both at the same time, even us with all our money, we cannot. Wall Street is not going to let us do that. They start by removing a lot of position. I think we see that accelerating, quite frankly. We have only seen the beginning, but in the next 2 years, we see a dramatic shift. But I think my position, I guess yours, and you know as well, is that there will be a lot more opportunities created as well, probably by also entities. Nuno Goncalves Pedro The Great Unbundling Yeah, there will be more opportunities created. The hoarding is just taken also a little bit of a different view. To your point, there’s hoarding of resources, compute, et cetera. But there’s also hoarding of top talent. We are seeing people getting paid, packages all in that could run up to 100 million, in some cases even over 100 million over several years. This is unheard of. I mean, an officer of Meta would make, I don’t know, maybe 20, 25 million a year. It’s like now there are people that are on the top end of AI researchers that are getting paid around that amount just to join some of these companies. There’s a little bit of a different hoarding. It’s very selective hoarding of certain talent. We’ve seen some acqui-hires. We’ve talked about it in previous episodes that are just literally about getting one or two people specifically to come on board. Alexander Wang, again, going to Meta to lead their intelligence labs there. I feel, I don’t know what you feel, but I feel this is a transition moment where there is overpaying for certain talent on the top of the market. At some point, this will stabilize. You can’t keep paying people 100 million over 4 years or something like that across the board. To your point, a lot of this is actually going to scale up quickly also on the AI side. There’s a little bit of a different hoarding happening on the top end, not just the resources, but also of people, which seems to give further this notion of barbell, that there’s two extremes, the haves and have-nots, the super-duper talented people that get paid a ton of money, tens of millions of dollars a year at the very least. Then the emptying of the middle where there’s a ton of tech layoffs going on in some ways, the belly, as they would call it, is being expelled. The middle market, the managers are being fired because there’s nothing to manage. There’s a lot of positions going away. In some cases, you might keep some of the more junior talent, but with a little bit of experience. But even the talent coming out of colleges is not getting hired either. It’s a little bit of a weird thing where there’s hoarding at the top, there’s an emptying of the belly, the middle, and then the early, early, early is also not getting recruited. It’s like what gives? How is this going to look in the future? I agree fully with you, Bertrand, that there’s a migration of this talent, not only to other companies, but also to other jobs. There will be new jobs that will emerge out of this. The DevOps, dev tools market didn’t exist until maybe 20 years ago at scale, and it got created. In some ways, we’re seeing there will be new markets, there will be new roles and new jobs that will be created around engineering teams going forward. We can’t anticipate all of them. But basically, the emptying of the belly is true as it’s happening right now. The low hiring on the early and the top end, getting tons of money. We think this is a transition to something else. There’s the hoarding of engineering in general is coming to an end at momentum. Now it’s time to rightsize teams, to get the right at the table, et cetera, and start figuring out what works and what doesn’t work. We’ve already had some horror stories coming out even from Amazon where they were breaking systems with their use of AI tools, and I’m sure it’s happening across the board. I’m on a board of a company and been tremendously affected by Meta and its algorithms, where basically because of advertising, there have been people served with ads for this specific company where the ad doesn’t match the company, so basic stuff like that. It’s been actually very, very difficult because in some ways, the company goes back to Meta. It’s like, “Hey, dudes, you guys are serving ads that are not even our ads with our copyright and stuff. How does this work?” They’re like, “Oh, it’s AI.” It’s like, “Well, it’s AI but can you give me my money back?” They’re like, “No, we won’t give you money back.” This creates huge issues for companies, for example, that are very dependent on advertising, which obviously there’s a lot of industries that are. They’re actually in production systems at scale. Meta is, I think now, the largest digital advertising in the world. I think they outgrew Google in one of the last quarters. Basically, this has a tremendous effect that systems that are in production at scale are getting inputs and changes driven by AI tooling, and somehow nobody can say what the hell is happening. Again, there will be a reckoning, there will be a redistribution, there will be a rightsizing of teams and an adequacy of teams going forward. I personally think this is a transition period. Bertrand Schmitt I think we are moving from hoarding or software engineering to hoarding the top of the top scientists in AI and hoarding of GPUs, GPUs/data center. For me, it was quite interesting to see the deal of Cursor with xAI, where basically they couldn’t get access to computing resources to run their model. But xAI had, I forgot the exact numbers, but close to half a million GPUs that no one, I mean, “no one was using” because their services are not so successful yet in terms of AI chatbot and the like. Basically, suddenly they are like, “You know what? We control access to resource.” But the new resource is, again, a mix of extremely talented AI engineering or AI scientists versus GPUs/data center. There is this race of controlling boss and everything else is going to be collateral damage. Some examples, I think, are quite interesting. You talk about some example of Amazon, even some production issues. I remember reading a quick post-mortem of one of the issues, and the conclusion was it was AI, definitely part of the issue. But the other part of the issue was AI used by junior engineers. For me, it’s interesting. It shows that actually junior plus AI is actually a danger zone. That’s why many companies are going to be way more careful. “Why do we need the junior people if they are just playing with fire?” I think we go back to that situation of barbell, as you call it. The top talents are extremely valuable because they know how a production system works. They are here to develop better AI systems. But the junior guys playing with fires, yeah, maybe it’s cute in startups, but in a big time production environment, a different story. Nuno Goncalves Pedro There will be a barbell with top-end talent super-mega paid and then mid-level talent that is individual contributors still doing a lot of great work, et cetera. Along the way, a lot of emptying of entry, a lot of emptying of the middle. Where does the talent go? The Three (?) Destinations I think we could say there’s three destinations for this talent. Maybe there’s four, maybe there’s more. Three that we can immediately identify. One is the AI native startup piece, where we have smaller teams that potentially get to a lot of revenue or top line over time, and where the Series Seed is the primary round, where we’re seeing Series Seed being raised of tens of millions of dollars, actually even hundreds of millions of dollars in Series Seed. In some ways, the stars there can get incredible compensations in terms of stock. They will stay for private and selling in secondaries later down the road because there’s so much capital at the table. Actually, in some ways, salaries are very high as well in some of these companies. It’s not like you’re trading off anything. You can get paid a lot of money. If your company at Series Seed for 10 or 15 employees has raised 50-$100 million, you can pay great salaries. In some ways, this is the extreme destination. The AI native startups that can make it is the extreme destination. Now, there aren’t a ton of AI native startups that can raise 50-100 million to 400 million in Series Seed, just to be clear. There’s a handful of hot deals in that space, but that’s one clear destination for top-end talent going through that. In that market, I think that’s one of the destinations. The second one is more what we would call the human-verified premium. It’s more of a play of companies that has still the need of human in the loop, either in terms of development, also in terms of activity, either because go-to markets are very intensive, and so therefore you need to have sales forces, partnership teams, et cetera. Or on the engineering side, it needs to have a lot of customization, integration. Companies are not just going to the, “Oh, you can come in and just apply your AI tooling and somehow magically the systems all work.” there needs to be quite a lot of and work and high touch work in getting stuff done. A significant part of that market, I’m not sure, is super VC investible. Maybe it’s a hybrid of private equity in VC, more PE style in many cases. It’s a PE-hold, sell to someone else market. As we’ve discussed in a previous episode on the SaaS-apocalypse, that hasn’t quite worked out for PEs. Question marks on how that human-verified premium market is going to evolve. But obviously, there’s a lot of work still to be done there, even on the engineering and science side. That’s the second potential destination. Then the third more aggressive destination is the reindustrialized middle companies that have a lot of specificity in going after small and medium businesses, local or regional affectations like ERPs or CRMs for specific markets, et cetera. Those are the three natural destinations. I would add the fourth, which is big tech. I mean, big tech doesn’t magically disappear, and I don’t think it fits neatly into any of these three markets. In some ways, big tech is now looking at the extreme for top talent a little bit like the AI native startup because they can pay. They can pay the 100 million every four years, et cetera. I do think it will typify taxonomically into a fourth type emerging, where, as we discussed, you’ll have top-end individual contributor talent. You’ll have the absolute top-end of the market because they can get paid. Then you’ll start having the emergence of earlier talent that is highly capable, et cetera. That will go back to a bit of a normal distribution in terms of talent on big tech. For me, those are the four destinations that I would put at the table. Bertrand Schmitt For me, big tech moving to big tech, I’m not sure if it’s really a destination. I mean, yes, in some ways it’s a reshuffle between the big tech companies. They are definitely all fighting in some ways for some of the same people. I can see that dramatic shift where big tech has to remove a lot of positions in order to replace by AI. Again, I think at this stage, it’s mostly driven by AI coding. We are still at the beginning because this is brand-new phenomenon that AI coding is so successful at its task. I don’t think it was true even 6 months ago. Some companies, take Anthropic, take OpenAI, are definitely there or close to be there in terms of no more writing of a single line of code by a human, zero. This is, again, 6, 12 months ago. Not true. But now it’s true in a few top companies. Take OpenClaw as well, most successful GitHub project of all time, not a single line written by its author. It would have been impossible. We’re talking about hundreds of thousands of line of code in a few months. It’s impossible to achieve that manually. If you look at the other big tech companies, the Google of the world, the Meta of the world, the Microsoft of the world, they are absolutely not there yet. They are going to be there because they have no choice. It’s you either go fast there or you die. You are not going to be able to survive competitors that are shipping 10, 50, 100 times faster than you are shipping. It’s a life and death situation. All the big tech companies are going to move, and mark my word, in the next 2 years from 10, 20% of AI-written code to 100%. During that transition, the next 2 years max, if you don’t do it in 2 years, you are going to die. Your stock price is going to crash. Then, of course, you will have to make changes. You will have to invest more in GPUs. You will have to invest less in your standard typical software engineer employees. Like you, I’m very optimistic that there are new buckets. AI-native startups definitely will be there. It will be transformational. Human-verified premium, very interesting category. In a way, it will be businesses that are inevitably less scalable through AI, and there is definitely a spot from there. I think the biggest would be the reindustrialized middle SMBs. Most of S&P 500 type of business are going to dramatically offer new software opportunities, new opportunity story to talented software employees because they will need to implement AI in everything they do. They will do it. They will need people who have software engineering knowledge in order to implement these systems. For them, what’s changing dramatically really is that thanks to much cheaper cost as thanks to AI coding, a lot of software projects that they couldn’t afford to do, that they couldn’t imagine doing by themselves, they are able to do it. They will invest in a lot more software capabilities than ever before. That will be a big game changer. And software, very tuned to their business model. There might be less buying of your traditional off-the-shelf SAF software and a lot more investment in a highly custom software by their own team, assisted with AI. I think that would be the part that is most transformed by all of this in a positive way. Nuno Goncalves Pedro Alternative Cap Tables, Alternative Compensation Models This will lead to a very fundamental shift, right back to the broken contract. What does the new contract look like? It looks like alternative cap tables depending on which bucket are you transitioning into. If you’re going into your AI-native bucket, and you’re a top-end talent, you’re like, “Dude, I’m worth 100 million over 4 years, so just compensate me accordingly with a mix of options in the company plus my salary.” If you’re top 1%, you can probably get away with salaries that you’d get anyway at mid-level from 300K, 400K and above, and you can get actually a lot of options already in the company. A lot of this is happening right now. There’s a premium for AI, we know that. There’s a premium for AI at the top end of AI researching, in particular on companies that are doing hardcore research on staff AI engineers, so companies that require actual AI engineering. There is a premium that is significant. It could be as high as 18% over non-AI peers, and it widens actually with seniority, shockingly enough. This is more of an average than anything else. Now, for me, and it’s for debate, but the perspective is this extreme comp will need to compress at some point. There will still be the haves and have-nots paid much better than the have-nots, so to speak, but there will be a compression. The variance can’t be the variance we’re seeing today for absolute top-end talent. That said, there will be variants. We know that big tech for over a decade, decade and a half, for example, in the Bay Area, has been paying a lot of money for director and above levels that used to be the VPs, so a million, a million and a half a year, all in compensations. It’s not unheard of that this will actually increase after this stage. That said, I do think that the compensation extreme that we’re in will get diluted down the middle. It will actually come down at some point. It’s part of where we are today. As we know, it is still a bubble. Bertrand Schmitt Yeah, it’s an interesting point. I think it’s possible. At the same time, that compression coming 2, 3, 5 years. At the same time, we have examples where there is no such compression. Take the top sports players in the world, golfing, basketball, NBA players. There has not really been any compression at all. For me, it’s interesting. If you look at the big tech companies, each being one of this top NBA team, why would such compression happen? As long as they are competing against each other and generating plenty of cash, I think there will be some fair question. We will see. I don’t have a strong opinion, but for me, it’s not a total given. Nuno Goncalves Pedro For me, the shocking thing is the faster AI becomes better, the more that compression will happen, because at some point, it’s like, why do you need the top talent as well? I don’t know. It feels like you’re trying to evolve a system that’s there to replace you. It’s like, “Okay, I’m getting paid 100 million over the next 4 years”, and then you develop something that’s so good that replaces you. Thank you. That’s cool. Bertrand Schmitt That’s a total possibility, yes, because we are in that very unusual market where the game is to only replace yourself and people like yourself. At some point, it is a possibility, I guess this one. Right now, we’re talking about replacing your “average software talent”. In 2 years, could we absolutely replace the absolute best top experts in the world? Probably. I think it’s just that at some point we’ll be reaching the stage where we strictly have no control anymore on our AI systems because no human is able to challenge and understand what’s produced. It’s not just a question of scale anymore. We’re talking about a gap in IQ, basically. Nuno Goncalves Pedro Exactly. It will happen at some point in history. We don’t know exactly when. For the second bucket, the human-verified premium bucket, it’s difficult to see how an HVAC company or an HVAC roll-up of scale or a regional health care platform or high touch go-to-market, B2B, SaaS play, et cetera, for a vertical will compete. At the same end, they have to compete and they will compete. There will be more and more jobs, we believe, for engineering talent in these companies. They’ll have to be more and more AI-enabled themselves. The cash salaries will have to be competitive within the local markets, not necessarily with Silicon Valley. There will be potentially profit sharing and revenue sharing and actual dividends played at the table. The model there on the cap table needs to change a little bit, needs to be probably propped up more on salary and on some way of doing profit sharing or actually having dividends paid to employees and figuring out employee to equity in a more aggressive manner. This is the market that probably was already very attacked, so to speak, or let’s say, occupied by private equity firms. There are still obviously part of that model that would work well. There needs to be a fundamental shift, certainly on the quantum of salary compensation, dividend compensation, profit sharing, and all of that. Then last but not the least, obviously, we had the bucket around basically the reindustrialization of the middle, so everything else, which will take most of the belly that we were talking about. This is probably a poor analogy, the belly fat. It’s not belly fat, it’s people that were doing their jobs that now are getting disrupted. In some ways, that bucket will absorb a lot of that belly, will absorb a lot of talent. The small and medium businesses that Bertrand was saying will need to crucially become more AI, software-enabled by themselves, even with some core stuff and underpinnings that actually might not even require AI in terms of infrastructure platforms. There, you need to get properly paid. Again, how many people do you need in your engineering team if you’re a small business? Probably not a lot. It’s maybe you need one or two people and that’s it. They’ll need to be very nicely paid because they’re running the stuff in the rails. This is probably a market that over time, as AI gets more and more competent, will also be disrupted, but let’s not talk about the disruption to the disruption because otherwise, we’ll stay here the whole day, but certainly a market that has a lot of potential to shift and to absorb a lot of the moments that we’re seeing in terms of layoffs happening in the US in particular. Bertrand Schmitt This category was a category that historically could not compete with Silicon Valley salaries, could not attract the most talented engineers. It’s not a category that didn’t want to bring these people on board. It’s a category that just couldn’t afford to bring this talent on board, typically. I think it would be a dramatic shift for them when suddenly there are opportunities to hire these people. There is an opportunity to hire them at maybe more reasonable prices from this company’s perspective. You talk about small companies, the great thing is that there are millions of small companies at some point. I think things could be truly transformational. Of course, some of these engineers, software engineers, might decide to become entrepreneurs on their own. Solo entrepreneurs, small businesses, build their own, easier to build their own product to market so to serve other companies. I think there will be quite dramatic changes because not all companies will be disrupted by AI as much, but not every company will benefit from improving processes, improving software through AI. At least early on, you will need this human touch to make it work inside a business. Interestingly enough, I was hearing that some companies like IBM were hiring more younger people to do the work of going to the client, understand their needs, propose implementation plans. That forward deployed engineer, those positions, I think there will be more and more available. Nuno Goncalves Pedro Investor Landscape Fragmentation What happens to investor into the landscape? We already had an episode, the previous one, Episode 76, where we talked quite a lot about the big capital reset on the private equity and private reset, including venture capital. Just maybe to summarize, how does it align with the buckets that we’ve just been discussing? I think the AI-native bucket clearly is going to be the key bucket. There, we’re going to see two movements. One movement, which is the mega funds, as we discussed in the last episode, are no longer just VC funds. They’re really mostly multi-asset private equity funds, maybe even private equity hedge funds in some cases. Those funds will be all over the high-growth AI-native companies and will be pouring money into companies that are scaling really, really quickly. The early stage, so to speak, VCs, the actual VCs that will stay in the market will be the guys probably identifying the next big wave of AI-native companies. We’ve discussed that as well in the last episode, some research that we did at Chamaeleon that I shared in episode 76. We’ll see that as emerging. What happens to the second bucket, the bucket around human premium, human in the loop? Likely we’ll have more and more private equity capital going into it and the large-scale VC guys, the Thrives of the world, they’ve just announced Thrive Holdings, and others going after those markets as well. It’s trying to converge into the private equity market, which aligns with the point we made in the previous episode that the VC mega funds are no longer VC, that they are private equity, multi-asset class. They’re going after a bunch of things. There’s a conversion happening from VC into private equity. It was going to happen anyway because the private equity guys were coming into VC as well and the hedge funds were coming to VC as well. There’s a convergence in the middle of very, very large funds and large assets under management happening to go after some of these opportunities, certainly in Bucket B. Then this Bucket C, so to speak, the bucket of reindustrialization, as Bertrand was saying, very well, likely will be self-funded for a significant period of time. Will self-fund with their own cash flow. Doesn’t need to have a ton of capital intensity. Maybe you need one or two engineers to do stuff, but that’s it. You don’t need tons of capital. You didn’t need in the past, you won’t need it today. Not sure there’s going to be a fundamental shift to that market. Bertrand Schmitt Yes, I certainly, overall, agree with you. That last pocket, probably little change to the capital and capital structure. Again, I see that as the biggest opportunity for a lot of people who might be less needed by big tech and also top tech companies. What is sure for the first category, the high native startups? I would say more overall in the VC ecosystem, there is no space left for SaaS anymore. I think SaaS, as we used to know it, is dead in some ways in the sense that new pure SaaS software startup are definitely out. Existing ones that are critical to run your infrastructure, the Salesforce of the world, I think they’re in a decent spot. Actually, interestingly, they changed their pricing model to now sell to AI agents, not just per seat. There is a change in pricing there. But this day and age of funding a pure SaaS software startup through VC money, no way. VC money going to AI-native startups, AI-focused startups, to biotech, to deep tech, to defense tech, yes. SaaS as a fundable category early on, I think it’s over. Nuno Goncalves Pedro I’m a bit more nuanced as we shared in The SaaS Apocalypse episode. We can call it whatever we call. It’s applied AI is the new SaaS thing. Horizontal applied AI is the new horizontal SaaS or vertical applied AI is the new vertical SaaS. I agree in common with your point that very specific point solutions around SaaS will be disrupted by nature with all the easy stuff you can do today with AI. It will take a while. This is not something that’s going to happen this year. It’s going to happen over the next years. Maybe interesting to also talk about the exit markets. I think the IPO market, as we’ve also discussed in the past, there is, in my view, going to be a reopening of the IPO market, I think this year, probably later in the year, third or fourth quarter. The median time to IPO actually is going to be really weird because there’s going to be potentially some companies in the current landscape, bubble or no bubble, that are going to IPO, the OpenAIs of the world, Anthropics of the world, et cetera. There will be more and more aggression, I think, on M&A. Big tech has already shown it, that they want to buy into markets. Large non-tech companies have also started doing acquisitions in space. To prop up their IT teams, their engineering teams with this world that we’ve also discussed in previous episodes that I’m going to own my own engineering stack for now. As we see, that normally doesn’t withstand the test of time. At some point it will get unbundled and served by someone else. Then finally, the secondary market is very hot right now. Obviously, there’s heavy discounting on some areas, high premiums on others. The exit market, strangely enough, is going to be propped up, in my opinion, over the next year to 2 years, dramatically. Then we’ll see if there’s a big reckoning around the bubble that we are clearly in or not, if it’s a soft landing or hard landing. Definitely, there’s going to be a lot of exit paths over the next year to 2 years. Bertrand Schmitt Concerning the “bubble”, I have two perspectives on this. One is it’s a bubble in the sense that money is going to a lot of players and some players are going to blow it up. There will be a concentration of players at the end, like it usually happens. If you look at, for instance, long time ago, the railway revolution, there was that intense influx of capital. At the end of the day, there was a dramatic change in transportation in the US and a complete railway system put in place. Yes, some investors lost money, some companies went bankrupt, but the transformation was fully real. There were a lot of top leaders at the end of this revolution. The change after that only happened, we guess, post-World War II, with the construction of the highway system and the rise of airlines and plane transportation overall. Here I feel it’s similar in the sense that, yes, there is a lot of money going in. Some players are going to blow it. They will misuse the money in different ways, but that’s part of dynamic allocation of capital. Of course, you make mistakes. That’s what happens. At the same time, I feel it’s a similar level in the sense of this is a dramatic change in the US infrastructure. This buildup of AI data centers filled with GPUs, integrated at scale with some of the best software in the world and running it, supported by a dramatic shift in energy infrastructure. This is for me similar to the Railroad Revolution. Some players might not own the data center they build because they didn’t manage well their debt, they didn’t manage to run proper software. You know what? They will get acquired by somebody else. I think we are at this level of fundamental transformation. The fact that in a matter of maybe 2 years, the move from 0% of code written by AI to 100 % written by AI is an insane dramatic shift. Just to be clear, when you move from manually coded to AI coded, we’re talking about a 100X difference in terms of speed at similar, if not better level of quality. The shift is dramatic, and on top of it, you don’t pay salaries anymore to achieve that. You pay CapEx, and with GPUs and OpEx with electricity. It’s a very big shift, positive shift in business model. New unions, no management over it, AI working 24/7. Personally, I think for me, bubble has a bad connotation in the sense of it was all for a waste. I don’t think it’s all for a waste. I think we are witnessing a dramatic revolution of our lifetimes, quite frankly, bigger than SaaS, bigger than mobile. From my perspective, it’s exciting times. Nuno Goncalves Pedro Operator Playbook and Predictions Let’s move to if you are this person, what would you do in the future? Let’s start with two extremes and go from there. One is you’re non-tech, so you’re not an engineer, et cetera. You’re trying to figure out, how do I scale my activity? Maybe physical labor is where I want to go. It’s not, “Go west” anymore. Definitely not necessarily go west. You should go to, I guess, the states that have no sales tax with very cheap energy because that’s where the data centers are being built if you want to be in that market. Obviously, there’s a lot of stuff that needs to be done: HVAC, electricity work, et cetera. Don’t go west. Go low sales taxes, low cost of energy. That’s likely where the data centers are being built. You probably can just follow. There’s, I’m sure, some way for you to follow where the data centers are being built, but that’s next, I think on that extreme of the table. The other extreme of the table, let’s say you are super ambitious, maybe you’re no longer an engineer, but you’re a product manager in your prompt engineering. You could do prompt engineering all day long. You’re 28, 29-year-old superstar. What do you go and do? Likely either you start your own thing, start your own company because you’re so good at prompt engineering, you probably can do a lot of the code yourself, particularly if you have an engineering background, or you go and join very early an AI-native startup that you think has the chance of going through the roof, and you take a pretty good salary early on, a ton of upside on the company because guess what? Companies like that need product managers. They need people to figure out UX, UI. It’s not going to be, at least for now, yet AI figuring that out for you. Those are two extremes, just to give two of the extremes, like engineering, product management persona, and physical labor at the other extreme, non-tech, et cetera. Bertrand Schmitt In some ways, every software engineering job is going to become the equivalent of a software engineering manager or a product manager, because suddenly you don’t have to do the coding anymore. You’re managing AI that is coding for you. Either you start to have some manager hat, but we saw the humans, so it’s a very different type of manager, obviously, or you are going to be really an empowered product manager. You’re skipping the middleman. You’re skipping the traditional engineering organization because your engineering organization is AI running and doing the work for you. I still believe that it requires some serious skills. I don’t believe in the vibe coder type of value proposition. I don’t believe in the prompt engineer becoming suddenly super incredible, able to manage that. I still think it requires some serious chops to do the best from all of this and to do it in a safe and sane way. It’s very easy to have poor taste, make mistakes. I don’t know you, but keep reading these stories on the heads of companies who lost everything because of the AI agents. That deleted stuff in production, and they had no backups or the backups weren’t deleted as well. Crazy situation. You cannot run companies like this if you let your agents running wild. You could argue it’s the early days. I would argue it that that issues would be there for a while. You need to have some engineering discipline at core in the company running the business to make sure things don’t go sideways because it would be easy for things to go sideways. Nuno Goncalves Pedro I totally agree. If you’re thinking, Oh, should my kid go into science and engineering and computer science, et cetera? Absolutely, still, because of everything that Bertrand just said. You need to understand actually what code does and what technology does and what all of that does. That’s still a skill of the future. It’s not a skill of the past. In some ways, it’s still a skill of the future very much. Maybe let’s try two more extremes. Around the same level, the person that decided to do an AI native company bootstrapped initially, having difficulty raising a mega round, but could probably get away with raising a 2-3 million seed round, et cetera. Is that still viable? The answer is yes. There’s tremendous capital efficiency right now happening in the market still, 10 plus higher than if you were doing a SaaS company, and you were a founder in 2019 or something like that. That capital efficiency is going to reverberate. You can run a tighter team, smaller team. Actually, you don’t need that many salaries. If you’re a decent engineer as a founder or if you understand enough as a product manager to just generate that code, you can do a lot of stuff yourself, can bring in maybe one or two technical elements to the team early on as you would have done if you were bootstrapped anyway. There’s obviously a path for that. The other extreme is you’re in big tech, you’re level five, individual contributor, making a ton of money, or you were a manager, and you’re now out of a job, where do you go? You can go to a big company that is non-tech, S&P 500 company that’s non-tech, something like that. You join the company, you’ll probably get paid pretty well, maybe not as high as you were paid in big tech. There’s some stock at the table, but guess what? You’ll have probably more work-life balance than you ever did. That’s the trade-off. You’ll have a better job. On the upside, you can transform the company. You can help and be part of transforming a company from non-AI to AI-first or AI-enabled in the future, whatever BS that will look like in terms of the argumentation to the board. You can actually create tremendous productivity enhancements in a big non-tech company if you come with that background. Again, you’ll have certainly a better work-life balance, so not a bad deal, to be honest. Bertrand Schmitt Also, to be clear, I talk a lot about AI coding because it’s truly transformational. You could argue that it’s going to be self-improving. We are in the situation of a self-improving AI that keeps improving itself thanks to automated coding. It’s a dramatic, virtuous loop. Obviously, AI is also going to improve everything else. It’s going to improve your marketing, it’s going to improve your search process, it’s going to improve your DNA. Improvements will be everywhere. It’s just that right now we are at a point in the quote-unquote revolution where there is one clear piece of the puzzle that is moving faster than the rest. Nuno Goncalves Pedro Bertrand, the senior executives at non-tech don’t know anything about that. It could be just a great prompt engineer. That’s the only job you do. “I’m the chief marketing officer. I have someone below me that’s doing the whole work.” Nobody knows. Nobody’s the wiser, I guess. I’m being facetious, but not fully. Bertrand Schmitt Yeah. There would be a transition period where what you described happen. I want to say, going back to AI coding, I think that the part of AI that as of today has reached a stage of limited AGI. We have reached, from my perspective, a limited type of AGI for coding. If you take coding as a discipline today, I think we reach AGI. If you go beyond coding, that’s true. If we are talking about coding, leveraging the latest LLMs: OPUS 4.7, ChatGPT 5.5, combined with Claude Code, Codex, and OpenCode for harness, I think we’ve reached AGI in the context of coding. I’m not sure everyone fully realize that and the consequence of that. I think the rest is going to come as well. We are going to see that category by category, usually categories that are more scientific in nature, where you can replicate, where you can test easily, where you can create clear success. Metrics will be the “easiest” to follow in that direction of self-improvement. I just want to highlight that this part is truly transformational, the root cause of everything we’re talking about today. At the same time, it’s coming beyond coding. Nuno Goncalves Pedro I think it is true. There are a couple of markets where that might not hold true, which is maybe the final path. If you’re thinking of starting your own business in plumbing and in HVAC maintenance and installation, this is a pretty good time for the reasons we already said before. There’s a lot of buildup of data centers and all that stuff, but also for other reasons, because it’s an activity that won’t be disrupted by AI yet. You need them embodied AI. You need physicality to AI to do stuff like actually fixing pipes. Bertrand Schmitt Until Optimus replace you. Nuno Goncalves Pedro Yeah, but if we’re 3, 4 years out in terms of a lot of these optimizations that we’re talking about at the software layer, we’re 10 years plus out on embodied AI, right? Bertrand Schmitt Oh, yeah, it’s 10 years. Nuno Goncalves Pedro We’ll probably be optimistic as we speak. That’s a nice business. I’m thinking of starting to go into that market. If you guys are interested in listening to this, just reach out to me. What’s the angle? I think there’s a lot of stuff you can do in the buildup of some of these businesses, plumbing, HVAC, all sorts of maintenance. There are markets that are just totally messed up. Handyman market in the US is totally messed up. There’s a bunch of companies out there that try to go after it with marketplaces and stuff. I honestly just start something from scratch, a small business, and go from there. Bertrand Schmitt Yes. They’re an interesting middle. Think about accounting firms, consulting firms. I think they are not as easy to replace, but at the same time, there is no way on what they do is not going to be dramatically changed with AI. I don’t know if it’s 50, 80, 90% of the job, but this is changing quite dramatically, would be my expectation in the coming few years. Conclusion Thanks for listening episode 77 of Tech Deciphered about that great talent redistribution. As you heard it from us, we believe there is a dramatic change in play, enabled by AI coding, and that ultimately a lot of the big tech companies are changing their employee distribution, way more focused on the top talents and bringing more GPUs. As a result, we will see a change in their staffing. Some of this change will benefit AI-focused startups, but probably more likely will benefit the bigger SMBs, the S&P 500 companies of the world that will finally be able to bring inside and afford some of the talent that were in some ways trapped by the top 5, 10, 20 software companies of the world. Thank you, Nuno. Nuno Goncalves Pedro Thank you, Bertrand
