AI Markets: Deep Dive with a16z's David George
David George of a16z analyzes the AI market, revealing that AI companies grow 2.5x faster than SaaS counterparts with lower sales spend, driven by immense market pull. He discusses the rise of 'Model Busters' that defy growth expectations, the critical 'adapt or die' moment for incumbents, and the massive, fundamentally sound infrastructure buildout, highlighting that we are still in the early innings of a major technological cycle.
The current AI landscape is defined by unprecedented demand-side strength and a massive, ongoing infrastructure buildout. Despite the rapid progress, we are only at the very beginning of a 10-15 year product cycle that is fundamentally reshaping technology markets.
The AI Growth Anomaly: Faster, More Efficient Growth
A core takeaway from market data is that AI companies are exhibiting a growth and efficiency profile vastly different from the previous SaaS era.
- Accelerated Growth: After a slowdown in 2022-2024, 2025 has shown a reversal, with AI companies growing approximately 2.5 times faster than their non-AI counterparts. The top-performing AI companies saw a staggering 693% year-over-year growth.
- Demand-Driven, Not Sales-Driven: This hyper-growth is not fueled by higher spending. In fact, the fastest-growing AI companies spend less on sales and marketing than their SaaS predecessors. The growth is a direct result of immense end-customer demand and compelling products.
- A New Efficiency Benchmark: A key metric, Annual Recurring Revenue (ARR) per Full-Time Employee (FTE), illustrates this shift. While the rule of thumb for SaaS was around $400,000, top AI companies are operating at $500,000 to $1 million ARR per FTE. This reflects not just leaner operations but the sheer force of market pull requiring fewer resources to capture.
- Gross Margins as a “Badge of Honor”: AI companies often have slightly lower gross margins due to high inference costs. This is viewed positively, as it signifies that customers are actively using the core AI features. There’s a strong belief that these inference costs will decline over time.
The “Adapt or Die” Imperative
For companies that are not AI-native, the message is clear: a fundamental shift is required for survival.
You you need to adapt to the AI era or die.
This adaptation must happen on two fronts:
- Frontend Product: It’s not enough to attach a chatbot to an existing workflow. Companies must aggressively reimagine their products from the ground up with AI at the core, even if it means disrupting their own business.
- Backend Operations: Internal processes must be overhauled. This means fully deploying the latest coding models and AI tools across every function. The productivity gains are significant, with one founder reporting a 10-20x increase in development speed after equipping two engineers with unlimited AI coding tools. This leads to a profound mindset shift, captured by one CEO’s question for every new task:
Can I do it with electricity or do I need to do it with blood?
This era also brings an evolution in business models, moving along a spectrum from Licenses -> SaaS (Seat-based) -> Consumption (Usage-based) -> Outcome-based. While outcome-based models are still nascent, the shift from seats to consumption can be highly disruptive for incumbents.
Validating Success: From Engagement Metrics to Enterprise ROI
Sustaining this rapid growth requires more than just initial sales; it demands deep and continuous product engagement.
- Harvey (Legal AI): Users are spending about double the amount of time in the product compared to before, indicating that AI is highly effective for legal work and lawyers are deriving significant value.
- Abridge (Medical AI): As the user base has grown massively, user engagement has remained consistently high, proving the product’s value to new and existing users alike.
- Navan (Travel): As a pre-AI company that successfully adapted, Navan now handles 50% of complex user interactions (like travel changes) with AI, resulting in a 20 percentage point expansion in gross margins.
While Fortune 500 CEOs express eagerness to adopt AI, real-world implementation is often slowed by the challenges of change management. However, for those who successfully deploy it, the impact is immediate and substantial.
- Chime: Reduced support costs by 60%.
- Rocket Mortgage: Saved 1.1 million hours in underwriting, translating to $40 million in annual run-rate savings.
The Macro View: A Fundamentally Sound Buildout
The public markets are being driven by AI winners, but this rally is built on solid fundamentals, not froth.
- Profits, Not Hype: Unlike previous bubbles, the current market is rewarding earnings per share (EPS) growth, not just loss-making growth. Valuations are based on sound fundamentals.
- “Model Busters”: AI is poised to be the biggest “model buster” in decades—a technology whose growth trajectory is so fast and long that it defies all conventional financial models, much like the iPhone did.
- A Healthy Capex Cycle: The massive infrastructure buildout is being financed primarily by the cash flows of historically profitable hyperscalers. This is a crucial distinction from past bubbles. For the ecosystem, this is excellent news as it increases the supply for training and inference. As one analyst noted, “there is no dark GPU”—unlike the fiber buildout of the dot-com era, every piece of AI hardware is utilized immediately upon deployment.
- The Path to Payback: To achieve a 10% hurdle rate on the nearly $5 trillion of projected cumulative capex by 2030, the industry would need to generate about $1 trillion in annual AI revenue. While today’s market is around $50 billion, it’s growing well over 100% year-over-year.