A Single Point of Failure Threatens Every Small Firm

When a handful of companies carry the weight of national security and of countless small businesses besides, every founder inherits a risk nobody chose on purpose.

A single point of failure is rarely the result of bad luck; it is usually the end point of years of convenient decisions, each one reasonable on its own. That is the real story behind the rise of one entrepreneur’s sprawling reach into rocket launches, satellite communications and now artificial intelligence infrastructure that the United States government depends on for core security functions. The remarkable part is not that one person built so much capability; it is that so many public institutions chose, year after year, to lean on the same supplier rather than cultivate two or three capable ones. Microentrepreneurs and startup founders should not read that as a story about Washington. They should read it as a mirror.

Concentration is seductive because it is efficient. One vendor, one contract, one integration, one invoice — it is always cheaper to manage in the short run than three vendors with three onboarding processes and three sets of quirks. The bill for that convenience only arrives later, when the single supplier raises prices, changes its terms, suffers an outage, or simply decides that a customer’s priorities no longer match its own. Governments can absorb that shock, slowly and at public expense. A five-person company usually cannot.

Look at your own stack with the same scrutiny. If your checkout, your customer data, your website hosting and your AI assistant all sit behind one provider, you have built exactly the structure now drawing scrutiny in Washington, just smaller and with less leverage to negotiate. The fix is not to reject powerful platforms — most are genuinely useful — but to make sure no single one can take your business down if it stumbles. A useful starting discipline is to map every tool a business cannot operate without for a week, then ask whether a credible second option exists; our rundown of practical AI tools for entrepreneurs is a reasonable place to start building that second option rather than defaulting to whichever platform is loudest this quarter.

The deeper lesson is about limits, not reassurance. A supplier that insists it would never abuse its position is making a promise; a market structure that makes abuse costly or impossible is a limit. The same argument applies to the way the artificial intelligence industry talks about its own power — a point we have made before in arguing that safety has to be built from limits, not promises. Founders betting their operations on any AI vendor’s good intentions are making the same mistake regulators have made with critical infrastructure: trusting character where contracts and alternatives should do the work.

Software work is consolidating in a similar direction. A shrinking number of agent platforms now sit between a small business and the customers, suppliers and code it depends on, deciding what gets automated and on whose terms. We have written about how that shift is changing jobs rather than eliminating them, and the same caution applies here: a founder who hands an entire workflow to one agent platform has handed that platform a veto over the business.

None of this is only a warning. Concentration this visible also opens a door. Customers who are nervous about depending on one supplier are actively looking for a credible second one, and that is a genuine opening for smaller, newer companies — provided they offer something distinct rather than a thinner copy of the incumbent. We have argued elsewhere that lookalike brands fail precisely when they try to win on sameness; the startups that benefit from this moment of scrutiny will be the ones that compete on reliability, transparency or price rather than on being a cheaper version of the giant everyone is suddenly worried about.

The practical task for any microentrepreneur or startup founder this quarter is an honest audit: list every function the business could not survive losing for a week — hosting, payments, communications, AI inference — and write down what happens if that single supplier fails, changes its price, or changes its mind. Where there is no real second option, that is a single point of failure, and it should be treated as a line item on the risk register, not a point of pride about how streamlined the operation looks on a slide. Build the backup before it is needed. Negotiate contracts that let you leave. Favour open standards over elegant lock-in. The businesses that do this now will be the ones still standing the day a critical supplier, however capable, has a bad one.

A single point of failure is rarely the result of bad luck; it is usually the end point of years of convenient decisions, each one reasonable on its own.

Build the backup before it is needed.

Sources

• How Musk gained so much power over U.S. security interests — PBS News Hour — Starting point for this editorial's discussion of vendor concentration risk for small businesses