Startup Statistics (2026)

Business creation in the US is running near record highs. Americans filed a record 5.67 million business applications in 2025, per the Census Bureau's Business Formation Statistics, above the prior record of 5.47 million in 2023 and well above the 2005-2025 annual average of about 3.46 million. The pace quickened in 2026: 4.93 million applications in January through September, up 15.3% on the same months of 2025 (see the table below).

Applications jumped from 3.5 million in 2019 to 4.4 million in 2020, then to 5.4 million in 2021, and have stayed above 5 million every year since. Not every application becomes an employer business, so these figures are a leading indicator of formation rather than a count of surviving companies.

The pandemic did not kill entrepreneurship, it accelerated it. Business applications surged about 24% from 2019 to 2020 and have never returned to pre-pandemic levels, a structural shift visible in five straight years of 5 million-plus filings from 2021 to 2025, with 2026 running ahead of 2025 (Census BFS).

Drivers include remote work, cheaper software, the creator and gig economy, and a wave of side businesses. The share of applications judged likely to become payroll employers has held up, suggesting the boom is not purely sole proprietors filing paperwork.

The real story is not how many start, but how many last. BLS Business Employment Dynamics data show about 77.9% of new US establishments survive their first year, 51.4% reach year five, and just 34.7% are still operating after ten years (see the chart and table below).

In plain terms, roughly one in five closes within twelve months and about half are gone within five years. Survival is steepest early: the curve falls fastest in the first two to three years, then flattens as the businesses that make it prove more durable.

You have probably heard that 90% of startups fail. That figure circulates endlessly in blog posts and pitch decks but is not grounded in reliable data, as CB Insights and BLS analysts both note. The government survival tables put five-year failure closer to 49%, not 90%.

The confusion comes from mixing definitions: venture-backed startups swinging for outsized outcomes fail more often than the average corner business, and "failure" can mean shutdown, acqui-hire, or simply not returning a fund. For the broad population of new businesses, roughly half survive five years.

When startups do fail, the reasons rhyme. CB Insights' analysis of 431 failed venture-backed companies found running out of capital at the top (70%), followed by poor product-market fit (43%), bad timing (29%), and unsustainable unit economics (19%) (see the table below).

Running out of cash is almost always the final cause of death, not the root problem: most companies that ran dry did so because they could not win or keep enough customers, which points back to product-market fit. An earlier CB Insights study of 110 post-mortems put "no market need" at the top (42%).

Where you start matters. Among industries, Information has the highest first-year failure rate at 28.4%, followed by professional and technical services (25.5%), while agriculture (14.3%) and accommodation and food services (14.7%) are among the most durable in year one (see the table below).

Geography matters too. First-year failure rates run highest in the District of Columbia (32.9%), Tennessee (29.3%), and Delaware (27.2%), and lowest in Washington (17.5%) and South Carolina (17.7%), per BLS data compiled by LendingTree (LendingTree).

Only a sliver of startups raise venture capital, but that sliver drives outsized innovation and wealth. US venture deal value reached $515.8 billion across an estimated 14,199 deals in the first three quarters of 2026, already the highest total in PitchBook-NVCA's dataset and about 44% above 2021's full-year $357.9 billion. Q1 and Q2 2026 were the two biggest quarters on record; Q3 cooled to $98.4 billion, down roughly 40%, even as its estimated 5,012 deals were the second-most of any quarter (see the chart and table below).

The path was volatile: funding fell from $357.9 billion in 2021 to $166.0 billion in 2023 as rates rose, then rebounded to $215.3 billion in 2024 and $319.7 billion in 2025. These are PitchBook-NVCA vendor estimates and are revised each quarter (the Q4 2025 edition had put 2025 at $339.4 billion).

The rebound has been almost entirely artificial intelligence. AI startups captured 82.7% of US venture deal value through Q3 2026, the highest annual share in PitchBook-NVCA's dataset, up from about 66% in 2025, 47% in 2024 and roughly 10% in 2015. AI's share of quarterly deal value fell within 2026 as the frontier-lab megadeals of the first half faded. In 2025, five companies, OpenAI, CoreWeave, xAI, Anthropic, and Databricks, raised nearly $60 billion between them (Q4 2025 edition).

Concentration was extreme in 2025: 487 mega-deals were 3.2% of deal count but 67% of value, and California alone drew about $191 billion, roughly 60% of all US venture capital. Half of all venture dollars went into about 0.05% of deals, leaving most non-AI sectors starved (PitchBook-NVCA).

Most startups climb a ladder of rounds. In 2025, median pre-seed rounds ran about $750K to $1.5 million on post-money valuations of $4-6 million, and median seed rounds ran about $3 to $3.8 million on $12-20 million valuations, per Carta and PitchBook data (vendor figures, flagged).

Round sizes rebounded from the 2022-2023 trough but sat 30-50% below the 2021 peak outside of AI. Angels, micro-VCs, and accelerators like Y Combinator write the earliest checks, typically $25K to $750K each through SAFEs or convertible notes.

A "unicorn" is a private startup valued at $1 billion or more, a term coined in 2013 when they were rare. The CB Insights list now counts 1,447 of them worldwide, collectively worth about $7.4 trillion, with the US home to 829, or 57% (see the chart and table below). PitchBook-NVCA, which tracks US companies only, counts a record 992 active US unicorns worth $5.7 trillion, up 14.9% year over year, with 179 new ones minted in 2026 through September.

The US and China together account for about 69% of the CB Insights list; China has 170, the UK 72 and India 63. AI accounted for more than half of the new unicorns minted in 2025, and AI-native companies are reaching $1 billion valuations faster, in about six years versus the historical seven.

Funding only matters if investors eventually cash out through an IPO or acquisition. US VC exit value was $274.5 billion across about 1,580 exits in 2025, and a record $2.29 trillion across about 1,240 exits through Q3 2026, almost all of it from a handful of deals: SpaceX's $1.7 trillion listing in Q2 (the largest IPO ever), its acquisition of xAI, and its $60 billion all-stock purchase of Cursor in Q3 (PitchBook-NVCA).

Still, the IPO window has not reopened: only 18 VC-backed companies went public in Q3 2026, 12 of them in healthcare, and OpenAI has ruled out a 2026 listing. Acquisition value hit a decade high of $465.3 billion, while the aggregate value of US unicorns stands near $5.7 trillion, much of it still locked up privately.

Young firms punch far above their weight in job creation. Census research found that firms less than six years old accounted for about 11% of employment but 27% of gross job creation, and firms aged one to five years generate roughly two-thirds of net new jobs.

Across OECD economies the pattern is even sharper: young firms are about 20% of employment but create nearly half of all new jobs (OECD). Startups are volatile employers, but their churn is where a lot of the economy's dynamism comes from.

The startup world is a barbell: millions of new businesses with coin-flip survival odds on one end, and a handful of venture-backed unicorns capturing enormous value on the other. Most everyday investors cannot buy into private startups before an IPO, and the base rates, half fail within five years, are a reminder of how concentrated the winners are.

For public-market investors, the practical takeaway is that the outcomes are power-law, not average: a few names drive most of the returns. That argues for diversification and for owning the winners through broad indexes or thematic baskets rather than betting the outcome of any single young company.