Capital Is Getting Expensive, Intelligence Is Getting Cheap
The cost of capital is rising while the cost of intelligence is collapsing. Their crossing redefines how companies should fund growth.

Two prices are moving in opposite directions right now, and I think their crossing defines the next decade of business.
Why Money Is Expensive Again
The first is the price of money. Sovereign yields are climbing (France's 10-year topped 5.1% in October, its highest since 2002), liquidity is draining into government refinancing, and leveraged positions are being unwound. Even Big Tech, growing 20%+ a year, is being marked down simply because the capital behind it costs more. The era of near-free money is over.
How Fast the Cost of Intelligence Is Falling
The second is the price of intelligence. For the same capability, the cost of AI has been falling roughly tenfold a year. GPT-3.5-level performance went from $20 per million tokens in late 2022 to $0.07 two years later. Analysis, writing, research, qualification and follow-up used to require people, weeks and budget. Now they cost almost nothing per unit and run around the clock.
Why Capital Was the Default Answer to Growth
For fifteen years, the default answer to any growth problem was capital. You needed more pipeline, so you bought ads. You needed more output, so you hired. You needed more speed, so you raised. Money was the cheapest input, so we used it to substitute for everything else.
That substitution is now inverting. The cheapest input is no longer capital, it's intelligence. And whenever the relative price of inputs flips, the winning business model flips with it.
Who Wins When the Cheapest Input Changes?
The companies that lead next won't be the ones that can still raise. They'll be the ones that learned to replace capital with intelligence: growth engines where marginal cost falls as volume rises, instead of budgets that grow with every new customer.
A Practical Test for Your Growth Budget
The practical test is simple. For every euro you plan to spend on growth this year, ask: could a system do this instead of money? Could an agent find, qualify or convert this demand at a fraction of the cost?
If the answer is yes and you still reach for the budget, you're solving a 2026 problem with a 2015 toolkit.
Cheap money was a subsidy that someone else controlled. Cheap intelligence is a capability you can build.
Keep reading
- The World Economy Has No Clear Direction, and that Is the Real ProblemThe global economy is not collapsing but offers no clear direction. Fragmented supply chains, institutional drift, and social misalignment explain the prevailing disorientation.
- Gemini 3: The Future of Artificial Intelligence in 2025Gemini 3 is Google's most advanced AI model, marking a new era of intelligence with state-of-the-art reasoning, multimodal understanding, and grounded tool integration.
- Power Laws Are Replacing AveragesWe built an entire civilization around the concept of average. Bell curves. Normal distribution. That world is disappearing.
- Generalists Are Winning Again, and AI Explains WhyAI integrates complete processes, not micro-tasks. The professionals best placed to use it think in systems. Hyper-specialization is now a vulnerability.
