The record volume of new business filings is not thin. The old definition of “substance” simply stopped matching how companies actually scale.
In my experience as a small business owner, an EIN still takes an afternoon. Building something that lasts takes years. The gap between those two things is real, but it no longer means the same thing it did a decade ago.
Americans filed 5,671,836 business applications in 2025. That is a record, up roughly 92 percent from the 2.95 million filed in 2016. The annual total has stayed above 5 million every year since 2021. The high-propensity series, the applications Census flags as more likely to produce payroll, grew only about 39 percent over the same stretch and reached roughly 1.71 million. The high-propensity share of all filings therefore fell from about 41.6 percent in 2016 to roughly 30.1 percent in 2025.
Those numbers are accurate. The Census Bureau’s Business Formation Statistics correctly track EIN applications from IRS Form SS-4. The January 2026 methodology change that pulled internet-sales applications out of the high-propensity series, and restated the history, is reflected correctly. The math holds.
The conclusion that the boom is therefore “thin in substance” does not.
The Payroll Filter Is Measuring Yesterday’s Business
Census defines high-propensity applications by a short list of signals on the SS-4: planned W-2 wages, a first wage-paid date, corporate form, or certain labor-heavy industries. That filter worked when almost every real business eventually needed employees. It is a weaker proxy now.
A single founder with AI tools, cloud infrastructure, automated payments, and a modern SaaS stack can carry the workload that used to require a full department. Minimal headcount is often the point, not the failure mode. A $10 million consultancy or software firm that runs entirely on 1099 contractors and fractional executives is classified as non-high-propensity simply because it has zero planned W-2 wages. That classification tracks a form, not economic weight.
Solopreneurship is no longer residual noise. Census Nonemployer Statistics show tens of millions of businesses without paid employees generating well over a trillion dollars in annual receipts. Many of those operators produce consistent cash flow, serve real customers, and build real wealth. Treating the majority of non-payroll filings as paperwork that never becomes anything understates a structural shift.
The divergence between total applications and high-propensity applications is therefore expected. Total volume measures the expansion of formal economic participation. High-propensity applications still track the narrower set of filings that intend to hire traditional employees. Both series are useful. Only one of them used to be treated as the definition of legitimacy.
What the Elevated Floor Actually Shows
Three durable changes sit underneath the numbers.
Filing friction collapsed. An EIN is now the low-cost gateway to business banking, digital payments, and global operations. More people take the first step because the cost of trying dropped.
Income diversification rose. The post-2020 floor shift reflects millions of people deliberately choosing ownership, full-time or hybrid, over exclusive reliance on a single traditional employer. That preference has held.
Capital-light models scaled. Founders increasingly design around high margins, software leverage, and contractor networks rather than headcount growth. The old equation “substance equals payroll” no longer maps cleanly onto the highest-margin paths.
Measure | 2016 | 2025 | Change |
Total business applications | 2,948,338 | 5,671,836 | 92% |
High-propensity applications | ~1.23 million | ~1.71 million | 39% |
High-propensity share | 41.60% | 30.10% | –11.5 points |
Source: U.S. Census Bureau, Business Formation Statistics. High-propensity counts rounded.
The table is the cleanest way to see the gap. Volume nearly doubled. The payroll-oriented subset grew by roughly a third. The share fell every year of the series. That pattern is real. The interpretation that the missing share equals missing substance is the part that no longer holds.
What This Means If You Already Run a Business
Competitive density is higher. Most new entrants are lean and often solo, so they compete on price, speed, and availability more than on scale. Differentiation through unit economics, reliability, and customer outcomes remains the durable response.
Hiring pressure has not risen in lockstep with total applications. That can create windows to secure strong people before labor markets tighten further. Pricing power also deserves a second look. Many of the new filings carry almost no fixed cost and can quote below your floor without going broke. Competing on rate against them usually ends badly. Competing on reliability and speed usually does not.
More formations also mean more potential acquisition targets and more operators who already survived the early chaos. For anyone buying rather than starting, a decade of elevated filings is a pipeline.
What This Means If You Are About to File
Filing is still the cheapest step in the process, so treat it that way. An EIN and an LLC cost an afternoon. Getting to durable revenue, credit, and clean books takes years, and most applications in this dataset never make that full trip. Industry-specific failure rates remain a more useful planning input than any national formation trend.
Two habits still separate the filings that go somewhere. Clean entity hygiene comes first, because a business credit profile only builds on real operating history. Honest capital planning comes second, because most founders overestimate what outside funding will do for a business with no traction.
The start-versus-buy-versus-stay-employed comparison frames the choice better than any national chart.
How to Read the Next Release
Census publishes BFS monthly, roughly eleven days after the month ends. Skip the total first. Then look at the high-propensity line as a share of the total, the same way you would read a jobs report past its headline number. One figure tells you how many people filed paperwork. The other tells you how many of them plan to pay traditional employees.
Both numbers matter. Neither one is a complete definition of economic substance anymore.
The boom in business formation is real. The metrics we inherited from a higher-friction, higher-headcount era simply no longer capture the full range of what counts.
Related Reading & Next Steps
- Start a Business, Buy One, or Keep Your 9-to-5 in 2026?
- Business Failure Rates by Industry: The Riskiest & Safest Sectors
- How to Buy a Business Instead of Starting One (And Why the Math Favors It)
- Business Credit Scores: What They Are, How They Work, and How to Build Them
- Debt vs. Equity Financing: How to Choose the Right Capital for Your Business