Winning in the Deflationary Era of Software

We’re entering a deflationary era of software. The cost of building has plummeted, and the upside has never been higher. AI, automation, and modular infrastructure are reshaping how software is created, scaled, and monetized. The result is smaller teams, faster iteration, and higher revenue per employee than ever before. 

An engineering leader who previously led large teams at Airbnb recently shared with me that between 2015 and 2020, Airbnb more than doubled its engineering team but wrote the same amount of code. In contrast, over the past year at Notion—where that same engineering leader now works— the team has grown only modestly but shipped twice as much. We’re seeing the same pattern across our portfolio. Companies are hitting milestones faster delivering value sooner with fewer resources. The playbook for scale is being rewritten in real-time.

The fastest-growing companies today—Cursor, Bolt, ElevenLabs—aren’t just scaling; they’re raising significantly more than their competitors. It’s not either/or between capital efficiency and fundraising—it’s yes, and. The best companies won’t just raise and raise; they will extract more leverage from every dollar.

A lower cost of capital is a strategic weapon. For example, if customer acquisition costs are the same across a given space, the company with the lowest cost of capital wins—acquiring more customers, expanding faster, and compounding its advantage.

Scaling well in this market will become less about needs and more about supply and demand. The old playbook said: raise big, hire big, scale big. The new playbook will leverage more from capital, move fast with AI, and stay lean.

While scale used to mean adding headcount, the next generation of breakout software companies will scale by augmenting their teams with AI-powered developer tools that accelerate code generation, automate testing, and ensure reliability. Instead of throwing more engineers at a problem, they’ll use automation to move faster with fewer resources. The result will be tighter feedback loops, higher-quality code, and an engineering culture prioritizing impact over output.

Infrastructure and model developers will absorb as much capital as the market allows, pushing the boundaries of what’s possible in AI and compute. But in the application layer—companies building on top of this foundation—can run much leaner. With off-the-shelf AI models, modular infrastructure, and automation-first workflows, application startups can achieve breakout velocity with smaller, highly leveraged teams. The best founders will recognize this barbell in the market, focusing on user experience, distribution, and business model innovation rather than raw engineering scale.

Here’s where I’m already seeing this shift: 

1. Margins are widening. Companies that optimize for low infra costs and automation will outcompete those burning capital on headcount. But if software gets cheaper to build, it may also get cheaper to buy—keeping margins constant long term. The winners won’t just be the most efficient builders, but those who turn their cost advantage into market dominance and durable business models.

2. Moats are evolving. Headcount was once a moat, but today, the operating principles "users first" and "efficiency is leverage" ring in my ears when I evaluate new investments. A lower cost of capital is now a strategic weapon. 

With equal customer acquisition costs, those with the lowest cost of capital win, outpacing competitors and compounding their advantage. The strongest moats come from deploying the least resources with maximum impact.

3. Leveraged talent. Hiring ten average people was never a moat, but the impact of one extraordinary hire now delivers a more significant ROI. A single, high-caliber builder—armed with AI, automation, and modular infrastructure—can outperform an entire team from just a few years ago.

The companies that understand this shift will hire talent that obsesses over users, rapidly adapts to feedback, and wields AI-powered tools to augment themselves. That may mean we’re in a moment where founders should deploy outsized levels of ownership to truly the best talent. 

The “best” talent, in this case, is a new class of builders—individuals leveraging ChatGPT, Claude, Cursor, Graphite.dev, Jazz.tools, and others to compound their efforts, ship faster, and deliver unmatched value. 

“Big tech” will become a relic of the past. It’s "high-leverage" tech now. And the best businesses will be built by those who understand that first. The strongest moats won’t come from pure headcount but from talent who can wield AI, automation, and capital with precision. Margins will widen for those who optimize, while those burning cash on scale will struggle to keep up. In the end, the companies that win won’t just build faster—they’ll build smarter, turning efficiency into dominance.