This article says AI is changing the game: small teams can now create huge value, and having lots of employees can be a liability. Author Ravi Gupta argues that the new success metric is “magic per employee” — for example, Cursor reached $100M revenue with only 10 people. He urges companies to become “world-class reactors” who quickly adapt, rather than trying to predict the future. He highlights three names: Cursor (tiny team, fast growth), Microsoft (investing $100B in AI), and ServiceNow (CEO focused on customers, met 1,000 clients in 90 days).
Sequoia Capital partner, former Instacart COO and CFO Ravi Gupta discussed in a podcast the core argument of his article "AI or Die": traditional enterprise success metrics such as headcount and process adherence may become liabilities. The AI era is breaking the historical constraints on small team
At a Glance
Ravi Gupta (Sequoia Capital partner, former Instacart COO and CFO) puts forward a core thesis in his article "AI or Die": Traditional corporate success metrics such as headcount and process compliance may become liabilities. The AI era is breaking the historical constraints of small teams and creating unprecedented opportunities. He emphasizes that "magic per employee" and organizational agility will become the key metrics for measuring value creation. Companies should strive to be "world-class reactors" rather than "predictors," and he suggests that board members actively drive AI adoption.
Gupta cites Dario Amodei's metaphor — "a nation's genius in a data center" — and Sam Altman's assertion — "within ten years, everyone will be more powerful than any single person today" — as the starting point for his thinking. He describes a thought experiment: a group of developers gathers on a Friday evening, ranks companies by market cap, employee count, and NPS (Net Promoter Score), looks for the target of "largest company, most employees, lowest NPS," and then competes to see who can rebuild it faster.
"If AI is really as powerful as we think, I don't think this is crazy," Gupta says. He points out that even without considering the most extreme scenarios, anything that prevents you from embracing change — a large workforce, adherence to quarterly earnings commitments, unwillingness to adapt — becomes a massive constraint. "If things change as fast as I think they will, anything that reduces your agility is a huge problem."
Implication: Gupta believes companies should analyze each role by asking "Can AI do this now? Can it do it in six months?" and recommends focusing on AI applications that are "currently barely viable but too expensive" — because models will improve significantly and costs will drop to one-hundredth within 3–6 months.
Gupta points out that the proportion of time CEOs and leadership spend on things that matter to clients is "surprisingly small." He cites HubSpot's Dharmesh Shah's view—SMBs are no longer "small and medium-sized businesses," but "small and mighty businesses." He proposes that "magic per employee" will become the new standard: "Wow, you did that with just 10 people?" Such amazement will become the new status symbol.
Data point: Gupta gives an example of a friend who manages a 400-person organization. Last year, the friend had to lay off 20 people, and subsequently spent a great deal of time handling the layoff process, communicating the plan, and calming the team—all time that did not directly serve clients.
Structural analysis: He proposes the test: "If you could rehire a person, would you hire them?" He believes that companies should "enthusiastically re-underwrite" their own business: What do clients need? What is required to deliver it in the best, fastest, and cheapest way? Are we currently doing that? Nothing should be considered sacred, including pricing models (e.g., shifting from per-seat pricing to pricing based on task completion).
Derivation: Gupta suggests that CEOs conduct a "calendar audit"—how much time do they and their leadership team spend on things that matter to clients? How much time do engineers spend writing code? The lower this proportion, the greater the room for optimization.
Gupta quotes Duke University legendary coach Coach K: "I am not a world-class predictor, but I am a world-class reactor." When college basketball rules changed (players went from staying four years to leaving for the NBA after one year), Coach K could not predict that change, but once "the game on the court changed," he could play it exceptionally well.
Mechanism Breakdown: Gupta distinguishes between two types of difficulty: "hard difficulty" (e.g., creating a new mathematical theorem) and "soft difficulty" (e.g., going to the gym). Becoming a world-class reactor falls into the latter category — it requires discipline, perseverance, and choice, not new intelligence. He quotes Matt Kohler's maxim — "Our job is not to see the future, but to see the present very clearly" — and emphasizes that when the present is changing rapidly, those who can respond quickly to new information will hold a massive advantage.
Falsification Signal: Gupta warns that the anecdotal experience that "AI has not changed your life" is extremely dangerous. "If you now think 'these models are not that good,' that may be interesting at a cocktail party and consistent with people's anecdotal experiences, but it is extremely dangerous. You should go sit with those who are fully immersed in AI and let them tell you about the future."
Gupta suggests that board members exert influence by asking one good question: "In a world of accelerating change, how do you feel about the person leading your company?" If the answer is "either very good or very tense," then action is needed.
He quotes former Sequoia partner Doug Leone: "In a world of accelerating change, how do you feel about the person leading your company?" Gupta adds that if the CEO is afraid to embrace change for fear of the board, the board should explicitly state, "We want you to go for it and explore all of this."
Historical analogy: Gupta borrows the story of former NBA player Shane Battier — as a child, Battier had a poster in his room that read, 'Somewhere, someone is training while you are resting,' and this "ghost" drove him to keep striving. When Battier entered the NBA, he encountered the real ghost – Kobe Bryant. "That ghost is out there, you have a chance to play against it, but you have to embrace it and go for it, because that ghost is tough." Gupta believes that for today's CEOs, there is someone out there using AI to create the 'magic' you promised your customers, and that person is working while you are resting.
Extrapolation: Gupta warns that private companies now have a huge advantage because "you don't have to answer for past public commitments." He praises Microsoft CEO Satya Nadella — in the Q4 2022 earnings call, he was still discussing "enterprise metaverse," and now he has shifted to "we will spend $100 billion in capex, going all in on AI" — "He doesn't care what he said on that previous call; he only cares about what happens next."
| Position | Analyst Sentiment | Key Data |
|---|---|---|
| ServiceNow | Bullish (customer-centric approach of CEO Bill McDermott) | Met 1,000 clients in the first 90 days, including weekends |
| Sierra | Bullish (AI-first, customer-oriented) | Founding team Brett and Clay; no financial data disclosed |
| Microsoft | Bullish (strong distribution + deep AI integration) | Plans $100 billion in capital expenditure (CapEx) |
| Instacart | Neutral (guest was former COO, believes its customer orientation is decent) | No specific data disclosed |
| Cursor | Bullish (small team + fast distribution) | $0 → $100 million in revenue, 10–30 person team |
| Cognition | Bullish (AI code agent) | Performs well due to shared code context |
1. “AI or Die” is not a threat, but an opportunity (Ravi Gupta): “In this moment, you can overtake the 15 cars ahead of you. If your company is worth $10 billion, it has a chance to become $100 billion; if it is worth $3 trillion, it has a chance to become $30 trillion.” Support: Scale constraints are broken, and small teams can create unprecedented value.
2. Become a world-class reactor, not a predictor (Ravi Gupta, quoting Coach K): “I am not a world-class predictor, but I am a world-class reactor.” Support: Predicting the future requires high intelligence, but reacting quickly to new information only requires discipline and choice—a capability entirely within your control.
3. “Magic per capita” replaces “how many people you manage” as the new status symbol (Ravi Gupta, quoting Dharmesh Shah): “SMB no longer stands for small and medium-sized business, but small and mighty business.” Support: A 10-person team generating $100 million in revenue at Cursor is more impressive than a 10,000-person team achieving the same revenue at a traditional company.
4. “Passionately re-underwrite” your business (Ravi Gupta): “You cannot treat anything as sacred—including your pricing model.” Support: Shifting from per-seat pricing to task-completion pricing may reduce revenue by 60% in the short term, but over the long term it can pursue a customer base five times larger.
5. Treat AI as “a genius colleague you fought tooth and nail to get” (Ravi Gupta): “If the first attempt is not perfect, you wouldn’t say ‘He’s no good’—you would say ‘I need to find a different way to bring out his full potential.’” Support: People currently hold AI to an extremely harsh standard, when failure often stems from their own prompts and insufficient context.
6. “Ghost” competition (Ravi Gupta, quoting Shane Battier): “Somewhere, someone is working on the ‘magic’ you promised your clients while you rest—and you will eventually face him.” Support: The “ghost” on Battier’s room poster eventually became Kobe Bryant; today’s CEOs must realize that competitors are using AI to build better products.
7. Don’t outsource your conviction (Ravi Gupta): “If you have a ‘convenient belief’ about something important, you should examine it seriously.” Support: Believing that “you can gradually transition to AI” is a convenient but dangerous belief; you should spend a month, with an open mind, trying to break your own preconceptions before making a decision.
8. Slope, not intercept (Ravi Gupta, quoting colleague Pat Grady): “For the best companies, the slope can be completely different—much steeper.” Support: This means being willing to pay a higher price for the best early-stage companies, but only if you find founders with ambition, curiosity, resilience, adaptability, and imagination.