Conference Presentation, Fireside Chat, Earnings Call
AI Markets: Deep Dive with a16z's David George
- AI demand is described as "extremely encouraging" with a product cycle at the "very beginning" of a "10, 15 year cycles," driving accelerated revenue growth in 2025 after a slowdown from 2022 to 2024.
- AI-native companies are reaching "$100 million bucks of revenue significantly faster" than historical SaaS counterparts, with top performers growing "693% year over year" while spending less on sales and marketing.
- Gross margins are expected to face pressure due to high inference costs, which are viewed as a "badge of honor" that will decrease over time; conversely, "super high" gross margins may indicate AI features are not being actively used.
- Operational efficiency is shifting, with best-in-class AI companies running at "$500,000 to $1 million per FTE" and product rebuilding proceeding "between 10 and 20x faster" using AI coding tools.
- A structural "turning point on code" occurred in December, with efficiency gains expected to take hold within the "next 12 months," creating a competitive gap where slower competitors will lag.
- Business models are predicted to evolve from seat-based to consumption-based, and eventually to "outcome-based" systems, with "early signs" of companies running "wholesale run[ning]... totally differently" expected over the "next five years."
- Specific company expectations include Navon maintaining a "20 percentage point expansion of gross margins" due to AI handling "50% of user interactions," and Flock clearing "almost 10%... more crimes" per officer.
- Enterprise adoption faces "change management" challenges despite strong CEO interest, with a "reckoning over the next five years" where non-adapting companies face a "huge disadvantage."
- Market dynamics indicate a shift toward private markets, with 86% of "$100 million plus revenue" companies now private, and value concentration in the "ten largest" unicorns doubling to "almost 40 percent" since 2020.
- Capital expenditure expectations project cumulative hyperscaler CapEx reaching "a little less than $5 trillion by 2030," requiring annual AI revenue of "about $1 trillion by 2030" to meet a 10% hurdle rate, with payback periods "between 2030 and 2040."
- Revenue-to-CapEx ratios are expected to improve "much faster" than the 10-year Azure timeline, with hyperscaler demand remaining "well in excess of supply" and older GPU generations maintaining "100% utilization."
- Financial projections estimate AI will drive "$9 trillion of revenue" and "$35 trillion of new market cap," with OpenAI and Anthropic projected to add "75 to 80 percent" of the public software industry's new revenue in 2026 against a current run rate of the "50 billion range."
- Supply side conditions are becoming "stretched a little bit" with no "dark GPUs" remaining, prompting the team to monitor debt in CapEx financing for risks associated with cashflow-negative ventures like Oracle.
- Public market earnings multiples are currently "higher than average" but below the dot com era, priced on "earnings growth," with a potential "ton more anecdotes" of productivity gains expected in the "next 12 months."