# When One Customer Powers Most of Your Revenue: What Robotics Companies Can Teach Startups About Anchor Accounts
A warehouse robotics company reported $720.8 million in quarterly revenue and $55 million in net income, yet a single customer and its affiliates accounted for 90.5% of that total. The company also recorded $19.4 million in investment gains that quarter, and its free cash flow ran negative. That mix of strong headline numbers and underlying complexity tells a story worth unpacking — especially for founders thinking about signing their own anchor customer.
## Why Dependence Can Be a Shield
The dominant customer in this story already wields significant leverage just by virtue of its size. When one account can single-handedly decide whether a company is profitable, the power dynamic shifts. The robotics company’s contracts with that customer include continuity rights — meaning the retailer can embed its own engineers into the company’s teams and access technical materials held in escrow under certain conditions.
A second agreement, finalized in early 2025, repeated many of those protections. It also imposed a cap — at a redacted figure — on how many comparable systems the robotics company could sell to each outside customer per year. The dominant customer contributed $520 million in development funding when that deal closed. In return, the robotics company accepted a $200 million base price for its robotics business, plus two potential milestone payments of $175 million each. Those milestones depend on hitting performance targets, so they are not guaranteed.
These terms read as preparation rather than a proven fallback. There is no public record of either side ever invoking those rights. But the mere existence of such clauses changes the bargaining dynamic entirely.
## The Hidden Cost of Every New Deployment
Robotics companies that rely on hands-on field work tend to grow their headcount alongside their customer count. At the end of its most recent fiscal year, the company employed roughly 2,000 people, about half of whom worked at customer sites installing, commissioning, and maintaining systems. Contractors supplemented the full-time staff when needed.
The company’s standard practice is to operate a new system for a limited period before transferring day-to-day management to the customer. This means field teams are shared between fresh deployments and ongoing maintenance. On paper, adding a second site might look straightforward. In reality, the same small team often carries both projects — and one that requires a different integration pulls even more people away from the primary rollout.
## How Long It Takes a New Customer to Go Live
Management has publicly stated a goal of adding one or two new customers per year. The historical timeline from signing a statement of work to system acceptance runs about 24 months. Some real examples illustrate this pace:
– A healthcare supply company signed a deal in late fiscal 2025 and plans a pilot in 2027.
– A wholesale grocery company announced a project in early 2026 and expects operations by late 2027.
– A food distributor signed an agreement in mid-2023 and had its first phase running by early 2025.
Revenue recognition can begin as costs are incurred once acceptance is reasonably assured, so the books may show activity well before a site is actually running. That gap between recognized revenue and real-world performance is where the real risk lives.
## Where New Demand Comes From
The company has broadened its customer base through ventures it helped create. A warehouse-as-a-service partnership with a major technology investment firm accounted for $11.6 billion of the company’s $22.5 billion backlog. The company owns a 35% stake and helps fund the build-out.
Another newer customer is a Mexican subsidiary of the dominant retailer, which carries an existing relationship into a new market. These structures make commercial sense, but they also mean some of the longer customer list reflects decisions made by partners rather than fully independent demand.
## The Scenario Most Planners Overlook
The most important scenario for founders to model is the quiet one: the dominant customer stays satisfied, meets all its commitments, and finishes its rollout on schedule. Once that program ends, the ongoing work shifts to service and software — and that typically brings in less revenue than installation did unless new system orders are already in the pipeline.
For the company in question, the distribution center rollout and its warehouse-as-a-service venture are the most likely candidates to replace that work. New accounts tend to start with a single site or a pilot, and management’s stated goal was one or two new customers per year. Even with revenue recognized early, a single new site adds relatively little compared to dozens of large projects moving simultaneously.
## Lessons Founders Should Take Away
### Scrutinize the contract, not just the check
When an anchor customer offers funding and scale in exchange for continuity rights and sales restrictions, the dollar amount deserves close examination — but so does every clause that limits where you can sell next. A restriction that blocks comparable sales to other customers for a set period has a real cost, measured in buyers you can no longer pursue.
### Test repeatability before scaling
The company has described moving testing to external suppliers and standardizing build and assembly procedures. That kind of work pays off on every subsequent deployment. Startups should track the same discipline: once a new site can be brought online with predictable engineering effort and field hours, the product is ready to scale.
### Start early with a second customer
A second customer on the same platform — even a small pilot — reveals how much effort a new deployment actually requires. The goal is not to generate significant revenue right away. The goal is to gather data on field hours, unplanned engineering time, and integration complexity.
### Model the quiet success case
Plans tend to focus on what could go wrong with the anchor customer. Far fewer models account for the dominant customer doing everything right — and then the rollout ending on schedule, leaving the company needing to replace that revenue stream.
## Cautionary Tales
History offers reminders of what happens when robotics companies lean too heavily on a single retail relationship without building independent resilience.
In one case, a shelf-scanning robotics spinout from a major university partnered with the same dominant retailer to place robots in about 500 stores. Plans to reach 1,000 stores were announced, but by late 2020 the retailer ended the relationship. The retailer’s leadership expressed concerns that customers did not perceive enough value from the robots to justify the cost. The robotics company laid off roughly half its staff, pivoted toward a different technology, and eventually ceased operations.
In another instance, an agricultural robotics company was acquired by a growing tech firm in 2021. When the parent company entered bankruptcy the following year, the robotics unit went down with it — despite its core technology having real-world potential. A separate agricultural robotics company shut down in 2021 even though its farming partner remained profitable. Its founder cited an inability to raise enough capital to complete development.
These stories share a common thread: a robotics company’s fate became inseparable from its customer’s fortunes or its own ability to secure independent funding.
## Contract Protections Are the Norm, Not the Exception
The company’s most recent annual report notes that the contracts behind its backlog lack termination-for-convenience clauses. Most of that backlog can only be terminated for missed performance standards, insolvency, or specified changes in control. This kind of protective structure is becoming increasingly common in automation contracts, especially as large buyers invest directly in robotics startups or deploy robots at scale.
For founders, the takeaway is clear: sort your anchor’s demand into what the contract commits the customer to and what the customer can still walk away from. Hiring ahead of expansion commitments means carrying those salaries if the orders never arrive — possibly for as long as it takes a new customer to go live.
—
## Frequently Asked Questions
**Q: Is it normal for a robotics company to have one customer represent over 90% of revenue?**
A: It is unusual, but it does happen when a single customer funds significant development and commits to a large rollout. What matters more is whether the contract includes protections, whether the work is repeatable, and whether the company can bring new customers online before the anchor’s program winds down.
**Q: How long does it typically take a new warehouse robotics customer to become operational?**
A: Based on publicly available timelines, expect roughly 24 months from signing a statement of work to system acceptance. Some companies aim to add only one or two new customers per year because of this slow cycle. Revenue recognition may begin earlier, but actual site operations take considerably longer.
**Q: What should a startup look for in an anchor customer contract?**
A: Beyond the price, examine continuity rights, sales restrictions, termination clauses, and what happens if the customer’s needs change. Every restriction on future sales has a cost. Every dollar of funding comes with obligations. Clause-by-clause review is far more valuable than focusing on the headline number alone.
**Q: Why do field hours matter more than revenue when evaluating a new customer?**
A: In robotics, deploying a new site requires engineers on the ground — for installation, troubleshooting, and ongoing support. A new customer pulls those same people away from existing projects. If field hours are unpredictable, growth becomes unreliable regardless of how promising the new logo looks.
**Q: What is the “no-villain” scenario and why is it important?**
A: The no-villain scenario is when the anchor customer does everything right, meets all commitments, and finishes its rollout on schedule. This is the outcome most plans fail to account for because it looks like pure success. But if the anchor’s program ends and has not been replaced, the company’s growth stalls despite everyone doing their job.
**Q: Can a robotics startup survive if its anchor customer walks away?**
A: It is possible, but only if the startup has diversified its customer base, built repeatable processes, and not hired ahead of guaranteed work. Contractual protections — like milestone-based funding and performance-linked payments — reduce the risk. But no amount of contract language replaces independent demand.
—
## Conclusion
Heavy reliance on a single customer is not inherently fatal — it depends on the terms, the protections, and whether the company is building toward repeatability. The robotics company discussed here has navigated its dependence successfully for years, generating profit and growing its backlog through partnerships and new ventures. But its story also underscores a principle that applies across hardware and robotics: the real measure of a business is not the size of its anchor customer, but what happens when that customer’s program ends.
Startups should model both the failure case and the success case. They should read contracts with the same discipline they apply to pricing. And they should start testing independence long before they need it — because the companies that survive are the ones that make their product repeatable before their customer list becomes their vulnerability.
Thank you for reading
