We’re back on 20VC, with Harry, Rory from Scale and SaaStr’s Jason Lemkin. On GPT-5’s strategic deflation, the ruthless efficiency revolution, and why you literally don’t need half your company.
Bottom Line Up Front
The Bottom Line Up Front:
- Jason’s Rule: “If you haven’t grown because of AI, you’ve failed.” After 18 months since ChatGPT launched, if your B2B company hasn’t re-accelerated growth through AI by June 30th, 2025, you’re likely irrelevant. Oracle did it. Intercom did it. If you’re still “working on it,” you’re already dying.
- Rory’s Reality: “Large amounts of money don’t change people as much as they reveal what they really are.” The AI investment boom ($300-400B annually) may not yield economically rational returns, and we’re seeing the inevitable outcome of fewer, bigger winners reshaping venture entirely.
- Harry’s Humility: “I think we overestimate our ability to predict our winners.” None of his Fund I predicted winners became actual fund returners, fundamentally challenging reserve allocation models and forcing brutal honesty about early-stage investing.
GPT-5: The Great Deflation
The consensus was clear: GPT-5 underwhelmed. But that might be exactly what the industry needed.
“Underwhelming kind of took a little bit of the air out of these techno-optimist ‘we’re underway to AGI’ narratives,” noted one of the hosts. “We’re now at the ‘it’s really great software for doing business, let’s make it better’ stage of life.”
The real story isn’t the marginal improvement in capabilities—it’s the business implications. GPT-5 is 8-10x cheaper than premium alternatives, creating massive pressure on competitors like Anthropic. For companies like Cursor, this is transformational: “If I’m Cursor, this is the best damn thing that ever happened. We have a competitive product at one-quarter the price.”
The Strategic Takeaway: We’ve moved from the “pie in the sky AGI” phase to grinding it out on business fundamentals. That’s actually good news for operators who need to build sustainable businesses rather than chase headlines.
The Chrome Wars: $34.5B for a Browser?
Perplexity’s rumored $34.5B bid for Chrome sounds insane until you understand the chess game being played.
Chrome itself generates minimal revenue—Mozilla makes only ~$700M annually despite significant market share. But Chrome’s real value isn’t in direct monetization; it’s in distribution control. “The most valuable person for whom Chrome would be valuable, if it wasn’t owned by Google, would be Google,” because they could simply pay $20B annually for search placement, just like they do with Apple’s Safari.
For an AI company like Perplexity, owning Chrome would mean instant access to billions of users. “Would an AI engine competitor to ChatGPT absolutely kill to own the Chrome user base? Absolutely.”
The Reality Check: This isn’t about the technology—it’s about distribution in the age of AI. Control the browser, control the default AI experience.
The $3B N8N Deal: When Workflow Automation Meets AI
N8N’s journey from a $300M valuation to $3B (reportedly at 40M ARR, ending the year at 80M) shows how AI transforms entire categories overnight.
“When you go from automating workflows in a very deterministic way to actually getting more of the work done using AI, the value prop of your software goes way up,” explained Rory. “Instead of saying ‘we’re going to automate a little bit of work,’ now we’re saying ‘we’re going to literally do the work and you can get rid of all these people.'”
The explosion in workflow automation isn’t just about connecting apps anymore—it’s about AI-powered work execution. “With the explosion of applications we’re building and the explosion of things we want to connect with AI, it’s like an order of magnitude bigger.”
The Lesson: Track companies that could benefit from AI acceleration, then ask which founders “get it” fastest and can ship LLM-enabled versions immediately.
Datadog’s ~$200M OpenAI Problem (That’s Actually Not a Problem)
Datadog had their best quarter ever—$260M net new ARR—but the stock still dropped 10%. Why? In part, concentration risk from their ~$200M annual OpenAI contract. AI customers are voracious customers, but there are a handful of leaders.
“If OpenAI continues its trajectory, it could be $520M in a year’s time — if it’s not negotiated down,” noted Rory. The real insight: companies that can “co-attach” to AI infrastructure spending are capturing massive value.
“Almost the entire GDP growth is AI capex. So if you can co-attach even if you’re not AI capex, if you’re not Nvidia, you’re going to get a pop.”
The Framework: Make a list of everything needed to build AI infrastructure. Chips, data centers, power, monitoring, security—then invest in companies that provide those building blocks.
Palantir: The Impossible Growth Story
From 12% growth at $2B revenue in 2023 to 45% growth at $4B ARR today. This “has never happened” in enterprise software at scale.
The secret? Palantir positioned itself as “the AI solution for large corporates.” When Fortune 100 companies need AI implementations, they can’t give $10M projects to small startups. They need proven vendors who can “look the CEO or CFO in the eye and say ‘we’ve done 10 of these.'”
Their US commercial bookings were up 222% at $843M last quarter. “Big companies need to spend big initiatives with big vendors, and all the other big vendors are old and stodgy like IBM and Accenture.”
But here’s the kicker: CEO Alex Karp said when they’re 10x bigger (at $40B revenue), they’ll have 10% fewer employees than today. That’s the future of B2B companies—massive scale with minimal headcount.
The Reality: At 120x revenue, Palantir is pricing in perfection. But if they maintain 40-50% growth for five years, they’ll be valued similarly to Google today. The question is whether that level of execution is sustainable.
The Shopify Efficiency Revolution
The numbers are staggering: From peak employment of 11,600 in 2022, Shopify cut to 8,100 employees while growing revenue 91%. That’s 30% fewer people generating nearly double the revenue.
“Toby went into beast mode and destroyed the competition. Big Commerce doesn’t exist. WooCommerce doesn’t exist anymore.” The efficiency gains came from AI adoption across the organization, with Shopify now generating $1.3M revenue per employee.
This isn’t unique to Shopify. “You don’t need half your company. You literally don’t need half the people working at your company.”
The Hard Truth: The coddling era of 2020-2022 is over. “Take three jobs, work two days from home, life is easy”—that world no longer exists. The founders who win will be “ruthless” like Toby, Zuck, and Karp.
The Great Reckoning: Who Survives the AI Transition?
The conversation revealed a brutal reality about employment in the AI era. Companies are deploying AI tools that immediately expose unproductive employees. “Every single time [an AI sales tool] has been brought in, someone on the sales team has quit the first day. Quit every single time.”
The pattern is clear: AI knows products better than most human employees. “Have you ever talked to an SDR that even understands the product they sell? Any 22-year-old? Once in my career, have I talked to a 22-year-old SDR that knows the product better than me. It’s worse with AI.”
The Survival Guide: Ask yourself—are you actually irreplaceable at your company? Not just as a person, but in terms of unique value creation. If you can’t answer that clearly, start building those capabilities immediately.
The One-Person Billion-Dollar Company
Is it possible? While the literal interpretation seems far-fetched, the broader trend toward extreme operational leverage is real.
One example: SaaStr itself is on track for $20M revenue with just two employees and 10 AI tools. “It’s not all better, and it’s a bit lonely, but we would never go back.”
The future likely isn’t one-person companies, but rather 20-40 person billion-dollar companies. “A core of engineers, a bunch of AIs, and people are going to say ‘I don’t want 100 sales reps.'”
The Trade-off: More PLG, more self-serve, more AI orchestration—less human management. “If you can choose an AI, you’re going to choose an AI over an unreliable resource.”
Venture’s New Reality: Concentration and Capital Intensity
The data is stark: highest seed valuations ever, but fewer deals being done. More importantly, there’s unprecedented concentration in late-stage rounds.
“For the last couple of quarters, we’ve literally had to back out one or two deals because they just make the statistics so weird.” OpenAI’s $40B round was twice the entire addressable market for many B2B VCs.
This isn’t temporary. “As these companies stay private for longer, there’s going to continue to be this steady stream of fairly humongous later stage financings.”
The New Model: Growth investors need to see clear paths to 3-5x returns, but the power law still applies—one in ten becomes exceptional. The concentration of capital into AI infrastructure companies provides justification for mega-funds deploying billions into individual companies.
Key Takeaways
1. Operational Efficiency Is Non-Negotiable
- Shopify: Since 2022: 91% revenue growth, 30% fewer employees
- Palantir: Plans 10% fewer employees at 10x current size
- The “ruthless” founders are winning
2. AI Infrastructure Spending Drives Everything
- Almost entire GDP growth is AI capex
- Companies that “co-attach” to AI infrastructure capture massive value
- Datadog’s ~$200M OpenAI contract is a feature, not a bug
3. The Employment Reckoning Is Here
- AI tools immediately expose unproductive employees
- 23-30 year olds without specialized skills face the biggest disruption
- The new job: Chief Orchestration Officer managing AI systems
4. Venture Capital Is Concentrating
- Mega-rounds are the new normal for AI infrastructure
- Traditional diversification models don’t apply to AI capex
- Growth investors can deploy billions into single companies
5. Distribution Beats Technology
- Chrome’s value isn’t the browser—it’s access to billions of users
- Platform lock-in remains powerful even in the AI era
- Control the default experience, control the market
Most Quotable Moments
On Operational Efficiency (Jason Lemkin):
“You may not need half your company. Palantir and Shopify are proving it.”
On Venture Concentration (Harry):
“Index are making everyone feel like shit right now. In particular in Europe, Sequoia and Excel, Excel have to win otherwise the gap widens more and more and more.”
On Market Reality (Rory):
“My big aha is it’s like dealing with a deranged madman trying to estimate what the street will do. I spend no time on this. Utterly unknowable.”
On the Future of Work (Jason Lemkin):
“If you can choose an AI, you’re going to choose an AI over an unreliable resource.”
On Palantir’s Positioning (Rory):
“They can look the CEO or CFO of a Fortune 100 company in the eye and say, ‘We’ve done 10 of these. You give us the $10 million, we’ll get this puppy done.'”
