AI Markets: Deep Dive with a16z's David George
Speakers
David George
— Senior Research Analyst, Baird
Summary
a16z's David George discusses the state of AI and private technology markets. He argues AI demand is extremely strong, AI-native companies are growing much faster and leaner than prior software cohorts, and public-market fundamentals remain sound despite high valuations. He also highlights AI adoption examples, portfolio companies, the capex buildout, and concentration of value in top private winners.
- AI demand is driving faster revenue growth and higher ARR per employee than the SaaS era.
- Pre-AI software companies must adapt their products and operations or risk falling behind.
- Public markets are led by AI winners, with earnings-driven fundamentals and limited froth.
- The AI capex buildout is massive but mostly financed by profitable hyperscalers, with debt worth monitoring.
- Private markets are increasingly important as companies stay private longer and value concentrates in top unicorns.
- Portfolio examples include Harvey, Abridge, ElevenLabs, Navan, Flock Safety, and Databricks.
- Older GPUs and TPUs remain highly utilized, easing depreciation concerns.
- Oracle's cloud bet and rising CDS costs are a risk to monitor.
Ideas
AI demand supports long product cycle
Coding has seen the biggest AI uptake and fastest productivity gains. A founder assigned two AI-deep engineers to rebuild a product from scratch with Claude Code, Codex, and Cursor, estimated 10-20x faster progress, and said the resulting coding bills would force him to rethink the entire organization. David expects all product and engineering organizations to work this way within 12 months, and companies that do not roll out the latest coding models will move much slower.
Legacy software must adapt or die
Pre-AI software companies must adapt or die. They need to reimagine products natively with AI, not just bolt on chatbots, and rebuild back-end operations with coding models and AI tools. Change management is hard, and business-model shifts from seat-based SaaS to consumption and outcome-based pricing could disrupt incumbents. Those that fail will move much slower and be at a huge disadvantage.
Shopify leads in AI adoption
Shopify is the coolest public-market example of a company running itself differently with AI. CEO Toby Lütke has led from the top, put AI at the center of the business, and performance-manages the organization to use it; David sees this as only scratching the surface of what will happen over the next five years.
Harvey engagement supports sustainable revenue
Harvey's legal AI is seeing much higher engagement because reasoning models are well suited to lawyering. Users spend about double the time in the product versus before, retention and engagement are strong, and this supports sustainable revenue growth. AI is not reducing the number of lawyers but making them more efficient.
Abridge engagement supports sustainable growth
Abridge is a medical AI tool doctors rave about, described as a trusted deputy. As user growth has massively increased, engagement among new users has held steady and grown slightly, which is a key sign that revenue is sustainable and not fleeting.
ElevenLabs voice AI usage surging
Voice is becoming the centerpiece of many new AI tools and workflows. ElevenLabs is growing very fast, usage growth is staggering, and it is an example of an AI company running extremely efficiently.
Navan AI lifts travel gross margins
Navan was early to the AI shift and applies AI to complex travel booking and change workflows; AI now handles 50% of those user interactions. This has helped expand gross margins by 20 percentage points over three years, while old-school competitors have not adapted, leaving Navan with a major margin advantage.
Flock Safety provides compelling crime-solving ROI
Flock has the most compelling customer value proposition in the portfolio because its ROI is solving crime. It solves about 700,000 crimes per year, and where Flock is deployed officers clear almost 10% more crimes, showing exceptional impact plus a strong business and financial model.
Chime AI cut support costs
Chime reported reducing support costs by 60% using AI, a concrete example of AI-driven cost improvement in a non-AI business. This is a developing productivity and margin setup worth monitoring as companies that successfully adopt AI may gain a significant advantage.
Rocket AI saved underwriting hours
Rocket Mortgage reported saving 1.1 million underwriting hours, up 6x year-over-year, and $40 million of annual run-rate savings from AI. This is a concrete example of AI improving productivity and margin in a non-AI business and is worth monitoring.
S&P 500 fundamentals remain sound
AI winners are driving nearly 80% of the S&P 500's return. Public-market fundamentals are sound, recent performance is driven by EPS growth, froth is minimal, and earnings multiples are above average but nowhere near dot-com levels. Investors are paying for profits, not loss-making growth, and David is optimistic AI will flow through to earnings.
High growth and margins rewarded
The market is rewarding growth, especially high-growth/high-margin businesses; high-growth/low-margin companies are also rewarded if unit economics are good and they are scaling into margins. Low-growth/low-margin companies should not be rewarded and trade low. Even high-margin companies without growth face a tough environment because growth is the biggest driver of 5-10 year returns.
High growth and margins rewarded
The market is rewarding growth, especially high-growth/high-margin businesses; high-growth/low-margin companies are also rewarded if unit economics are good and they are scaling into margins. Low-growth/low-margin companies should not be rewarded and trade low. Even high-margin companies without growth face a tough environment because growth is the biggest driver of 5-10 year returns.
AI infrastructure buildout remains attractive
The AI capex buildout is massive and concentrated, which is inherently risky, but the underlying fundamentals do not resemble previous bubbles. It is financed primarily by highly profitable companies, paybacks on training models look good, and demand exceeds supply. David is all for adding as much capacity as possible for training and inference, while monitoring debt and counterparty quality.
Hyperscalers are strong capex counterparties
Leading tech companies are the best businesses in history, with strong margins and cash flows. The hyperscalers are bearing most of the AI capex, and David is comfortable with Meta, Microsoft, AWS, and Nvidia as counterparties because they are historically profitable and generate cash flow; not all counterparties are equally safe.
Oracle cloud bet raises credit risk
Oracle is making a bet-the-company move into cloud, committing very large amounts of capital and going cash flow negative for many years. Its credit default swap cost has risen to about 2% over three months, and David explicitly says not all counterparties are equally safe; this makes Oracle a riskier AI infrastructure counterparty.
AI model companies growing rapidly
The model companies are scaling at a staggering pace. OpenAI and Anthropic together added almost half of the net new revenue that the entire public software industry added in 2025, and in 2026 the AI model companies may add 75-80% as much as the entire public software industry. AI revenue is roughly $50B today but growing well above 100% YoY toward a potential $1T by 2030.
Top private tech value concentrates
Private markets have become a real asset class as companies stay private longer and 86% of $100M+ revenue companies are private. Value is concentrating in the largest private winners: the top 10 unicorns make up almost 40% of the $5.5T value of North American and European unicorns, a share that has doubled since 2020. David has portfolio coverage in seven of the top 10.
Databricks is an AI transition winner
Databricks has transitioned from a pre-AI company to an AI-embedded one: its data lake is a strong place to run AI workloads, it is aggressively iterating on new AI products like Agent Bricks, and cutting-edge AI-native companies are customers, validating its technology. CEO Ali Ghodsi combines commercial and technical depth, and Databricks has low-cost technology plus a chance to grow with its scaling customers.
Up Next
This a16z video, published February 09, 2026, features David George discussing AI-SECTOR, Pre-AI software companies, SHOP, Harvey, Abridge, ElevenLabs, NAVN, Flock Safety, CHYM, RKT, SPY, High-growth, high-margin companies, High-growth, low-margin companies, Low-growth, low-margin companies, AIQ, META, MSFT, AMZN, NVDA, ORCL, OPENAI, ANTHROPIC, Private technology companies, DATABRICKS. 19 trade ideas extracted by AI with direction and confidence scoring.
Speakers: David George · Tickers: AI-SECTOR, Pre-AI software companies, SHOP, Harvey, Abridge, ElevenLabs, NAVN, Flock Safety, CHYM, RKT, SPY, High-growth, high-margin companies, High-growth, low-margin companies, Low-growth, low-margin companies, AIQ, META, MSFT, AMZN, NVDA, ORCL, OPENAI, ANTHROPIC, Private technology companies, DATABRICKS