Should You Expand Capacity Based on a Single Client?

When one or two clients account for most of your revenue, every promise from that client to keep ordering more sends you into a fresh round of second-guessing: do you trust their word and scale up equipment and staff in advance, or do you hold off and go find other clients who haven't made any promises at all? Last week, Amazon's own moves offered a supersized version of exactly that dilemma.

What Happened

Amazon raised its capital expenditure guidance again, to $220 billion. The interesting part is that the main driver behind the increase wasn't a surge in demand — it was rising memory chip prices, in other words, higher costs. Over the same period, the company issued $25 billion in bonds to fund AI infrastructure, and its debt load nearly doubled in six months, from $65.6 billion to $128.9 billion. Operating cash flow remains strong, but investment is outpacing it, pushing free cash flow into negative territory.

None of this came out of nowhere. Long-term fiber-optic contracts, in-house AI chip development, an expanded European logistics network, and a $48 billion investment in India are all part of a pattern of pre-positioning infrastructure that has been building for several quarters. Over that same stretch, AWS posted its fastest growth rate in 18 quarters, proving the profit engine is still running strong.

The rationale behind this aggressive spending becomes clear once you look at what's backing it: confirmation that AI infrastructure demand from Amazon's top four customers will hold through 2028. Meanwhile, the pipeline of new customers is showing signs of stalling.

What This Move Reveals

Look at Amazon's profit structure and a single business — AWS — accounts for $39.8 billion in operating income, or 58% of the company's total profit. Nearly everything else it's expanding, from retail to its investment in India, is effectively riding on cash generated by this one engine. And that engine's revenue, in turn, leans on a small number of major customers.

Given the near-certain demand from existing customers, it's a rational calculation to pre-build infrastructure even if it means taking on debt. But the fact that debt doubled even as operating cash flow kept climbing signals a shift — from investing what you earn to investing what you borrow. The trouble is that this strategy compounds two risks at once. One is that when uncontrollable costs like memory prices rise, the investment burden balloons faster than planned. The other is that the more a company concentrates on its surest customers, the more customer concentration risk becomes baked into its structure.

It's in this same context that management laid out its vision of eventually growing AWS into a trillion-dollar business. But reaching that target will take more than expansion from existing large customers — the stalled new-customer pipeline will eventually need to be refilled. The moment demand looks most certain may be exactly the moment the customer base is at its narrowest.

The Lesson for Solo Entrepreneurs

Scale aside, the logic is the same. When a key client's orders start growing, our instinct is usually to lean in further — improve service for them, buy equipment on their timeline, push other sales efforts to the back burner. If even Amazon couldn't escape this pull and ended up with a stalled new-customer pipeline to show for it, solo entrepreneurs need to consciously do the opposite. If revenue is climbing but your list of new clients hasn't changed, that may not be growth — it may be deepening dependence.

Start by calculating what share of revenue your top client accounts for, every quarter. If one client accounts for more than half, that's a warning sign for customer concentration risk. It's safer to limit upfront investment to demand that's confirmed in writing — a contract, a deposit — and to cover anything based on a verbal promise with variable costs like outsourcing or leasing. The extra volume a client says they'll give you is still just a line in your revenue forecast, not revenue you actually have. And even in your busiest month, set aside fixed time to meet new clients — prospecting for new business isn't something you do when you have spare time; it's something you do precisely when things are going well.

One Thing to Try This Week

This week, pull the actual numbers: what share of your revenue over the past year came from your top two clients? If that figure is above 50%, customer concentration risk is no longer someone else's problem. Before deciding on your next equipment purchase or hire, the right first step is reaching out to three prospective new clients.