20VC: Applovin: $160BN Market Cap, $5.48BN Revenue, $10M EBITDA Per Head | Why the Best Do Not Need Mentorship | Why Founders Should Not Angel Invest | Why Kindness in Business Will Slow You Down with Adam Foroughi

What It Covers

Adam Foroughi, CEO of AppLovin ($160B market cap, $5.48B revenue, $10M EBITDA per employee), details how the company rebuilt its machine learning architecture during a 92% stock collapse in 2022, cut headcount by 50% during triple-digit growth, and constructed a lean team of ~400 people generating extraordinary financial results.

Key Questions Answered

• •Equity compensation structure: Limit equity grants to the top 10-15% of employees who can absorb stock volatility; pay everyone else in cash with optional ESPP participation. AppLovin caps stock-based compensation at roughly $300M annually against a $150B market cap, keeping dilution minimal. Evaluate companies on cash flow minus SBC, not EBITDA alone, to avoid being misled by companies buying back shares just to offset dilution.
• •Rebuilding technology under pressure: When AppLovin's stock fell 92% in 2022, Foroughi halted all R&D on the existing recommendation system model and rebuilt the entire architecture using current machine learning techniques. This required replacing personnel committed to the old system. The new model, Axon 2, launched April 2023 and drove near triple-digit revenue growth, with EBITDA margins reaching 84% and a Rule of 40 score of approximately 150.
• •Strategic buyback execution: Rather than buying shares on the open market, AppLovin identified specific cap table holders who needed liquidity—private market investors, ex-cofounders—and negotiated direct repurchases using operating cash flow plus raised debt. This eliminated concentrated selling pressure before new institutional investors entered. Foroughi estimates this approach generated roughly $50B in value, approximately one-third of the company's current market cap.
• •AI-native org design: AppLovin eliminated its product organization entirely, requiring engineers to own product decisions. Engineers must audit AI-generated code for security and quality rather than simply prompt agents. Approximately 90% of code involves AI generation, but the metric tracked is revenue generated per token spent, not code volume. Avoid token budgets or usage leaderboards—they incentivize output with no business value, mirroring the bloat created by headcount hiring quotas.
• •Lean team construction: AppLovin's core advertising business runs on roughly 400 people. HR was reduced from 70-80 people to approximately 15 individual contributors. The executive team consists of CEO, CTO, CFO, and General Counsel only—no CRO, COO, CMO, or CHRO. Foroughi's method: identify process-oriented roles, eliminate the processes, then remove the people who maintained them. Cutting 50% of a mediocre team leaves 50% mediocrity; rebuilding requires removing 99% and starting over.
• •Founder investing trade-off: Foroughi stopped angel investing entirely because generating liquidity to invest requires selling shares in AppLovin, which contradicts the goal of maximizing the core business over a 20-year horizon. Every hour diverted to evaluating external investments represents an unmeasurable but compounding loss to the primary business. For founders who believe their company has the highest risk-adjusted return available to them, diversification through angel investing is a distraction, not a hedge.