Oct 28, 2024 · 1h 16m · news
Daniel Khachab: "We Are in the Middle of a Cold War for AI Talent" | E1220 · 20VC with Harry Stebbings
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of the 20VC podcast, Choco CEO Daniel Khachab discusses his high-stakes pivot from a traditional SaaS model to AI agents, Europe's tech sovereignty crisis, and his intense philosophy on leadership and fundraising.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 19.4% of the talking time here. How this is scored →
Harry as informed peer 5.0 Guest teaching 4.6 Guest disagreement 3.1 Harry pushing back 3.6
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 11:06 Disagreeing on AI Adoption Speed in Traditional Industries
Daniel directly rejects Harry's premise that traditional European enterprises will take ten years to adopt AI, arguing AI eliminates user interface friction compared to legacy SaaS.
Hardest push from Harry ▶ 43:59 Challenging the 'Take the Cash' Venture Valuation MindsetHarry forcefully refuses Daniel's justification for raising at a $1.1B valuation, arguing that taking massive cash injections destroys execution speed and capital efficiency.
Biggest teaching moment ▶ 34:50 Debunking European Regulatory Excuses with Elon Musk ExampleDaniel educates European founders on stoicism, pointing out that Elon Musk built Giga Berlin in heavily regulated Germany while local founders make excuses about notarization.
Harry holds his own ▶ 3:44 Citing Proprietary Data Moats and Llama Open-Sourcing StrategyHarry demonstrates deep sector expertise by citing Choco's specific food distributor relationships, historical invoicing data, and Meta's open-source Llama strategy to challenge Daniel's claim that tech moats are dead.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Pivoting Choco from SaaS to AI | 2 | 2 | 1 | 1 | Harry introduces the episode by asking Daniel why Choco pivoted from a darling SaaS model to 100% AI revenue. Daniel collaboratively explains his sleepless nights worrying about code automation and how software moats could vanish overnight. | |
| The Debate on Moats: Proprietary Data vs. AI Tech | 7 | 4 | 4 | 7 | Harry aggressively defends Choco's position against Daniel's own pivot, citing specific data moats like distributor relationships, order histories, and Meta's open-source Llama strategy. Daniel concedes that partial moats remain but insists agentification is killing traditional SaaS UIs. | |
| Overcoming AI Adoption and Implementation Barriers | 7 | 6 | 6 | 7 | Harry brings enterprise statistics showing European legacy businesses barely know Slack or Notion, arguing agentic SaaS is a decade away. Daniel directly rejects the premise, arguing AI is easier for traditional industries than software because it requires no learning curve beyond prompting. | |
| Value Accrual and Foundation Model Dynamics: OpenAI vs. Anthropic | 5 | 4 | 2 | 3 | Harry probes where value accrues across the stack and asks Daniel to evaluate OpenAI versus Anthropic in a hypothetical buy scenario. Daniel breaks down Anthropic's computer use features and commoditization of model pricing. | |
| Labor Shortages, Job Replacement, and AI Reallocation | 6 | 5 | 5 | 6 | Harry raises Klarna's support layoffs and cites Adam Smith's invisible hand regarding wage adjustments for unwanted jobs. Daniel counters that state involvement in sectors like German childcare creates market failures that free markets cannot fix. | |
| The Reality of AI Layoffs: Choco's Hard Decisions | 5 | 3 | 2 | 4 | Harry presses Daniel on the internal human cost of Choco's AI transition. Daniel vulnerably admits to laying off hundreds of high performers due to AI automation in account management and customer care. | |
| The Global Cold War for AI Talent | 4 | 5 | 2 | 3 | Harry prompts Daniel to explain the 'AI talent cold war'. Daniel details dinners hosted by US, UK, and Gulf officials offering golden visas and state-subsidized salaries to lure AI founders. | |
| Europe's AI Sovereignty Crisis: Chips, Energy, and Models | 6 | 6 | 5 | 6 | Harry challenges why Choco's R&D is in Berlin rather than London's premier AI institutional hub (DeepMind, Oxford, Cambridge). Daniel defends Berlin for application-layer development while highlighting Europe's existential lack of chip manufacturing, energy, and foundation models. | |
| Rebuilding Europe's AI Pillars: Chips, Energy, and Models | 6 | 5 | 4 | 5 | Harry quotes Intercom's Des Traynor regarding European regulatory paralysis and argues Europe can build foundational models like Mistral. Daniel explains why misdirected state funding (Intel) and structural capital gaps handicap European foundation model ambition. | |
| Debunking European Excuses: Funding, Regulation, and Work Ethic | 5 | 6 | 6 | 3 | Daniel delivers a fiery critique of European founder whining, insisting capital is a flight away and regulation is a stoic obstacle to overcome, citing Elon Musk building Giga Berlin. Harry strongly concurs with Daniel's anti-excuses philosophy. | |
| Choco's Global Portfolio Strategy: US, UAE, and Europe | 4 | 5 | 2 | 2 | Daniel outlines Choco's global market portfolio strategy across the US, UAE, and Europe. He recounts the nightmare of losing 98% of Choco's GMV overnight during COVID and using extreme travel hacks to keep sales teams active. | |
| Valuation Regrets and the Unicorn Trap | 6 | 5 | 4 | 6 | Harry challenges the wisdom of taking high venture valuations, arguing bloated cash piles ruin capital efficiency. Daniel agrees with Harry's critique, calling the unicorn status a cultural trap that creates complacent employees looking for safety. | |
| COVID-19 Management Mistakes and the Importance of Purpose | 5 | 3 | 3 | 5 | Harry pushes Daniel on work-life balance, insisting that top-tier execution requires sacrificing home life presence. Daniel agrees that founders and teams must be permanently ready for war. | |
| Fundraising Secrets and Building Generational Companies in Europe | 5 | 5 | 2 | 2 | Daniel outlines his fundraising principles and demystifies VC psychology, noting investors fear missing iconic outlier deals far more than taking write-offs. Harry agrees on European founder pitch deficiencies. | |
| The Masochism of Entrepreneurship: Would You Do It Again? | 5 | 3 | 1 | 1 | Harry quotes Jensen Huang and Alex Will on founder masochism. Daniel agrees using a ski instructor metaphor about kids who fall down learning fastest. | |
| Strategic Mistakes: Skunkworks Teams and the Spanish Market | 4 | 5 | 2 | 2 | Daniel reflects on wasting money on 'skunkworks' special project teams rather than forcing core teams to innovate. He breaks down why Spain was Choco's hardest market due to extreme supplier fragmentation. | |
| Going All-In on Product-Market Fit and Respecting Competitors | 4 | 4 | 2 | 2 | Daniel reflects on the need to go all-in once product-market fit is proven. He describes fighting multi-frontier rival Rekki when they had ten times Choco's capital. | |
| Quick-Fire: AI-First Beliefs, Rapid GMV Growth, and Corporate Dysfunction | 4 | 5 | 3 | 2 | In the quick-fire segment, Daniel reveals Choco scaled to $2.5B GMV post-AI shift. He calls out Deutsche Bahn corporate dysfunction for paying full executive bonuses while train punctuality tanked. | |
| The Core Purpose of Tech: From 10-Person Unicorns to Generation 3 Impact | 5 | 6 | 3 | 2 | Harry and Daniel reflect on Silicon Valley's infrastructure decay despite trillon-dollar tech creation. Daniel introduces 'Generation 3' companies where profit directly correlates with environmental or social improvement. |
Statements from this episode (34)
Khachab: AI will replicate years of proprietary software development by 2028
“This thing is going to learn how to code in no time. And maybe not now, maybe not next year, maybe 27, maybe in 28, like where's our technological mode is going to be? Like it might just be gone, like anyone maybe able to replicate what we have built over year…”
Khachab: Tech workers without AI building skills will become obsolete
“If you were to leave Joko tomorrow and you would not have the skills On how to build a FAI, then your skillset will be obsolete.”
Stebbings: Meta open-sources Llama because its true value is proprietary data
“Why does Facebook open source Lama? Because the values in the data that they have. A hundred percent.”
Khachab: Future software companies will sell virtual AI employees, not features
“But in the future, we are more or less selling an employee. And like, you know, here's a very highly competent employee that is, that has specific skills, sorry, to perform in your company.”
Stebbings: 62% of large European enterprises do not know what Slack is
“In Europe, 62% of large enterprises still don't know what Slack is.”
Khachab: AI adoption will be faster in traditional industries than in tech
“So the adoption curve is going to be way faster. Like AI is the perfect tool for traditional industries. Much more actually than for kind of like startups and tech, because those are people that are tech first, they know how to work with interfaces and stuff l…”
Khachab: Choco's AI API costs dropped 80% in six months despite growth
“We literally had to do nothing, and our price is now 80% lower than six months ago, despite probably having five or six extra transactions.”
Khachab: Anthropic is a better investment than OpenAI due to higher revenue
“So I'm not sure if these numbers are right, but I recently read that Anthropic is more or less doing 60% of the revenue of OpenAI, and they do this with less customers, so they kind of achieve a higher price point, probably. Plus the recent release of computer…”
Stebbings: Klarna CEO Sebastian Siemiatkowski fired 700 customer support employees
“Seb at Clon has fired 700 people in customer support.”
Khachab: Germany lacks 50,000 kindergarten workers despite billions in state funding
“Like just in Germany, we're short 50 K. 50 K and the government has invested billions over the last, we're still short 50 K.”
Khachab: AI will not enable enterprises to replace Salesforce with internal tools
“I don't think so. Because, okay, like AI today, like even the best LLM won't enable you to build something like Salesforce fairly quickly. There's so much code, so much customization in there. And then plus, even if it would, then you would have to maintain it…”
Khachab: Choco automated large parts of customer care and account management
“We also automate a large part of our customer care, large part of our kind of more account management can kind of work and replace it with AI.”
Khachab: Foreign governments offer to pay top AI talent salaries for years
“Like the most extreme offers that you get is like people are going to pay Sell it like government is going to pay salaries for three to five years. For your top AI talent. Like, they're gonna do that.”
Khachab: Europe lacks AI sovereignty because it misses chips, energy, and models
“In Europe today, We're not producing the chips. We're not producing the energy. We don't have the foundational layer models. So in that sense, these countries that don't possess those three elements are not really sovereign countries when it comes to AI.”
Khachab: Europe will receive AI products last due to regulatory friction
“I'm not sure if we're not gonna get it, I think we're just gonna get it last. It's gonna just, it's just gonna take longer.”
Khachab: Europe should subsidize TSMC and NVIDIA instead of Intel
“First of all, Intel is the wrong company to fund. So the first step you ask me is like, put it to the right company. Put it to Taiwan Semiconductor. Put it to NVIDIA. People that can actually produce GPUs.”
Khachab: Europe should not rely on nuclear power for AI energy needs
“I'm personally not, not, not a big fan of nuclear, just because I think it puts a lot of responsibility on future generations for problems that we haven't solved yet.”
Stebbings: European founder quality is much worse than in the US
“I think the quality of European founders is much worse than the US”
Khachab: No German VC fund could lead a €20M round in 2018
“When we started Choco, there was no way To raise twenty million, even just twenty million euros in Germany. It just did not exist. That kind of funds that you could raise one, you could raise five, maybe 10, 20 just did not exist.”
Stebbings: Europe has too much VC capital for too few good founders
“First, I think we do have enough domestic capital now, like the amount of incredible and not incredible funds in Europe. There are too many. I think there's too much money for two little good entrepreneurs.”
Khachab: European regulation does not hold back most SaaS or marketplace startups
“Look, there's certain heavily regulated industries, and I don't know them very well, but I'm sure eventually they're at a disadvantage. But that doesn't hold, for example, for most of SaaS, and also neither for most of marketplaces.”
Khachab: Investors should short Europe under the current government status quo
“So I think status quo, we have to be short. Government is just, yeah.”
Khachab: Choco lost 98% of its GMV in the first COVID week
“When COVID hit and when you lose 98% of your GMV at, and it's a very first week of it”
Khachab: Reaching a $1B valuation weakens startup urgency and attracts safe players
“Now the real downside to me is that it's at one billion mark, because once we become unicorn and people think, oh, we've made it. Like we've done, we, we're great now. Kind of people get less hungry. And then you get applications of people who look for a safe …”
Khachab: Giving employees a week off during COVID was a mistake
“First week of COVID, everyone gets a week off. And I said, okay, it might be great. Everyone can adjust the new situation. Like the opposite was the case. Like purpose was gone. So I think that was a mistake.”
Khachab: Skunkworks teams are a flawed approach to driving company innovation
“And I think today I'm convinced that that that's the wrong way to think about company building. Actually this AI would be another example. Like we could give it like to a special project kind of skunk works kind of team. You figure out AI and then you give it …”
Khachab: Spanish restaurants order from 18 suppliers compared to four in US
“The average restaurant in the US and Germany orders from four and a half different suppliers. Yeah. So say one for fish and one for meat and one for drinks and so on. France we had nine, much more specialty, more focused on food and so on and so forth. In Spai…”
Khachab: Choco should have tripled its burn rate in its first years
“I wish you would have, you know, spent double the amount of money, have tripled the amount of burn probably in the first 24 months of our company.”
Khachab: Only companies with majority AI revenue will win in the market
“I do believe only AI first companies will win. And by AI first, I mean that a majority of your total revenue or of your new revenue, which will lead into a majority of total revenue, Has to come from an AI product.”
Khachab: Choco added $1B in GMV in under a year post-AI transition
“I can tell you our GMV, since this AI transition grew like tremendously, I think it took us with kind of like the two years of COVID break, it took us probably five years to get to one and a half billion-ish, and it took us then less than yet at another billio…”
Khachab: France and Italy ban German trains due to extreme delays
“French and I believe even Italian railway system doesn't let German trains anymore enter their countries because they're just too fucking late all the time.”
Khachab: AI automation will enable 10-person unicorn companies
“I think one thing that became very clear is that, you know, great to me, great companies, at least from headcount perspective, get smaller and not bigger. And the reason is, is like you have to automate as much as possible and with AI that is fairly possible. …”
Khachab: Eliminating food waste has 5x the CO2 impact of eliminating cars
“Reducing food waste. And I'm not saying, you know, alternative foods. I'm not saying eat less meat. I'm not, I'm just saying don't produce that waste. That's the only thing. Has five X the impact as if all cars would be gone, not electrified, but gone. Five X …”
Shorts cut from this episode
▶ Why this CEO didn't want to celebrate his company hitting $1 (@0:21)
▶ Is San Francisco broken? 🏚️ · 20VC with Harry Stebbings (@1:11:02)
▶ POV: AI has just taken your job 😱 · 20VC with Harry Stebbin (@20:45)
▶ Are we in an AI Cold War? 🤖 · 20VC with Harry Stebbings (@23:14)
▶ Is AI bringing an end to SaaS? 🤖 · 20VC with Harry Stebbing (@0:05)