Startups get talked about as if they were one thing. In practice a startup is simply a young company built to grow fast, usually by selling something that did not exist in that form before. That is a small definition, but the effect on the wider economy is not small: the products these companies ship end up setting what customers expect from everyone else, including the established firms they compete with.
This guide looks at what startups are actually doing to business trends right now, using the data that can be checked rather than the stories that get repeated. How many new companies are being formed, how many survive, where the money is going, which sectors are genuinely being reshaped, and what all of that means if you run a small business, lead a team, or are thinking about founding something yourself.
Key Takeaways
- US business formation is running at over half a million applications a month, but only about a quarter of those look likely to hire staff.
- Global venture funding hit a record $510 billion in the first half of 2026, and two AI labs took 43% of it.
- Roughly 22% of new US businesses close within a year and 49% within five, so the widely quoted “90% fail” figure overstates the risk.
- AI is no longer a category investors screen for. It is assumed, which raises the bar for what counts as a real advantage.
- Exits reopened in 2026 after several quiet years, which changes the calculation for founders and early employees.
How Many Startups Are Actually Being Founded?
Before looking at what startups change, it helps to know how many there are. Two US government datasets answer that, and they tell a more grounded story than the funding headlines do.
What the Formation Data Shows
The US Census Bureau publishes monthly Business Formation Statistics, which count applications for a new employer identification number. In July 2026 there were 578,926 such applications after seasonal adjustment, 8.1% more than in June. That is a high level by historical standards and it has stayed high since the pandemic-era surge.
The more useful number sits inside it. The Census Bureau separates out “high-propensity” applications, meaning those with characteristics that historically predict the business will actually hire employees. In July 2026 there were 151,857 of those, roughly a quarter of the total. The rest are mostly sole traders, side projects and holding entities. So the headline figure describes enthusiasm for self-employment; the smaller figure describes the pipeline of companies that might one day have payroll.
That distinction matters for anyone reading formation data as a growth signal. A record month for applications does not mean a record month for new employers. If you are curious about the smaller end of this spectrum, our piece on micro-entrepreneurship and one-person businesses covers how that segment behaves differently from venture-backed startups.
How Many Survive
The “9 out of 10 startups fail” line is repeated constantly and it is not what the data says. Using Bureau of Labor Statistics business survival figures, an analysis by LendingTree found that 22.1% of new US private-sector businesses close within their first year, and 48.6% have closed by year five.
Failure is still the single most likely outcome over a decade, but the shape of the risk is different from the folklore. Most new businesses survive their first year. Half survive five. The 90% figure usually comes from venture-backed startups specifically, a much narrower and much riskier group, and applying it to all new businesses inflates the danger of simply starting something.
The Funding Picture: A Record Total That Hides a Narrow Market
Startup funding in 2026 looks spectacular in aggregate and difficult up close. Both things are true at once, and the gap between them is the single most important fact about the current market.
Where the Money Went
According to Crunchbase, global startup investment reached $510 billion in the first half of 2026. That is a record for any half-year period, and it exceeds the $440 billion raised across the whole of 2025.
Then look at how it was distributed. OpenAI and Anthropic together accounted for $217 billion, or 43% of everything raised in those six months. More than 70% of second-quarter capital went to companies working on AI. In the second quarter alone, 16 companies raised rounds of a billion dollars or more, worth $108.6 billion between them, which was 53% of the quarter’s total. US-based companies took 67% of second-quarter funding.
In other words, a handful of very large companies raised sums that distort every average in the market. This is not a rising tide. It is a small number of very deep pools.
What This Means If You Are Not an AI Lab
For a founder raising a seed or Series A round, the record headline is close to irrelevant. The money concentrating at the top does not trickle down to smaller rounds, and it makes the funding environment harder to read. Investors comparing your metrics against a market where the biggest deals are frontier AI labs are working with a distorted reference point.
Three practical consequences follow. First, evidence of revenue matters more than it did in 2021, because capital is available but selective. Second, the pitch that “we use AI” no longer distinguishes anything, a point we return to below. Third, capital efficiency has become a positioning tool rather than just a constraint, which is why so many companies now lead with a path to profitability rather than growth alone. For a deeper look at how rounds are being structured, see our overview of current startup funding trends.
Exits Reopened
The other significant change in 2026 is that companies started getting out again. In the second quarter, 32 venture-backed companies went public at valuations above $1 billion, and 24 companies were acquired at $1 billion or more, together worth $113 billion.
Exits matter beyond the founders involved. When acquisitions and public listings stall, capital stays locked in existing portfolios and investors have less to recycle into new companies. A reopened exit market feeds the next funding cycle. It also drives consolidation, which we cover in our analysis of how mergers are reshaping the SaaS industry.
The Sectors Startups Are Reshaping
The trends worth tracking are the ones where startup activity is changing what the rest of the market has to offer. Four stand out.
AI Has Stopped Being a Category
The clearest signal of how far AI has spread is that specialists have stopped counting it separately. Rock Health, which tracks digital health funding, discontinued its “AI-enabled” label because AI had become common enough to be meaningless as a distinguishing feature.
Y Combinator’s Winter 2026 batch shows the same pattern from the other direction. Of 199 companies, 56 (28%) build AI-native services and 45 (22%) build AI-enhanced software. Half the batch, before counting developer infrastructure and AI research companies on top.
The practical implication for founders and buyers alike: AI is now assumed, so the interesting question has moved. It is no longer “does this use AI” but “what does this company have that AI alone cannot supply”. Usually the answer is proprietary data, a regulated license, a distribution channel, or a workflow customers have already built their operations around. Our guide to industry-specific AI solutions covers where that defensibility tends to come from, and how AI is changing business operations looks at the buyer’s side of the same question.
Health Technology
US digital health startups raised $7.4 billion across 244 deals in the first half of 2026, up from $6.4 billion over roughly the same number of deals a year earlier, according to Rock Health. The median deal was $14 million, up from $12 million in 2025.
The concentration pattern repeats here in miniature: 19 companies raised 20 rounds of $100 million or more, accounting for 45% of all capital in the sector. Larger cheques to fewer companies, rather than a broad-based recovery. If you work in or sell to healthcare, our pieces on healthtech business opportunities and digital transformation in healthcare go into what is actually being bought.
Climate and Energy
Climate tech is the sector where honest reporting requires a caveat. Sightline Climate recorded $40.5 billion of global venture and growth capital into climate tech in 2025, up 8% on 2024. Comparable full-year figures for 2026 are not yet published, so treat any 2026 total you see as provisional.
What is visible is a shift in emphasis. Rising electricity demand, driven in part by AI data centers, has pulled investment towards power generation, grid infrastructure and energy efficiency, and away from the consumer-facing climate products that attracted attention several years ago. Our overview of climate tech trends tracks where that capital is landing.
Fintech and Embedded Finance
Financial technology has settled into a quieter, more infrastructural phase. The consumer app boom has given way to companies that sell payment, lending and compliance capability to other businesses, which then present it under their own brand. That is what “embedded finance” means in practice: your accounting software offering you a loan, or your marketplace paying out to sellers without a separate bank.
The business logic is straightforward. Selling infrastructure to businesses produces steadier revenue than competing for consumers, and it puts the startup in a position where switching costs are high. See our coverage of fintech trends and embedded finance for how these arrangements are structured.
How Startups Changed What Customers Expect
The most durable effect startups have on business trends is not the companies themselves. It is the standard they set for everyone else.
Think about what you now assume when you buy something online: that you can see where your order is, cancel without phoning anyone, get a refund without arguing, and reach support in a chat window rather than a queue. None of that was standard fifteen years ago. Startups introduced those expectations by making them a selling point, and once enough customers had experienced them, they stopped being differentiators and became the baseline that established companies had to meet.
The same thing is happening now with personalization. When a store remembers your sizes, or a service anticipates what you are likely to need next, the underlying technology is unremarkable. What it changes is your patience for companies that do not do it. Our article on e-commerce personalization covers how far this has gone, and the piece on direct-to-consumer brands shows how selling straight to customers changed retail pricing and returns policies.
There is a limit worth naming. Customers reward convenience and consistency far more reliably than they reward novelty. A startup that wins on a genuinely better experience tends to keep its customers; one that wins on being new tends not to.
The Hard Parts of Building a Startup in 2026
The current environment has specific difficulties, and they are not the ones from the 2022 downturn.
Capital is available but concentrated. Raising is not impossible, but the bar is higher and the comparison set is skewed by enormous AI rounds. Companies that would have raised comfortably in 2021 now need working revenue.
Technical differentiation erodes quickly. When the same foundation models are available to everyone, a clever product built on top can be matched within months. Durable advantages now come from data, regulation, distribution or entrenchment in a customer’s workflow, not from the model itself.
Compliance arrived earlier than most founders planned for. The EU AI Act’s transparency obligations came into force in August 2026, which means companies deploying AI systems that interact with people in the EU have disclosure duties regardless of size. Rules for higher-risk uses were pushed back, but the transparency layer applies now. If your product touches EU users, this is a build requirement rather than a later legal task.
Cheap building has raised the noise floor. The same tools that let a two-person team ship a product also let a hundred other teams ship something similar. Being first to build is worth less than it was; being first to reach customers is worth more.
None of this argues against starting a company. It argues for starting one where you have an unfair advantage that is not purely technical, and for a plan that assumes you fund growth from revenue longer than you would like.
Collaboration, Open Source and Cheap Infrastructure
One structural change genuinely favors small companies: the cost of the underlying technology has collapsed.
Open-source software lets a startup use databases, machine learning frameworks and developer tooling that would once have required enterprise licenses. Cloud infrastructure means you rent capacity instead of buying servers. Model APIs mean you can add capable language features without training anything yourself. A team of three can now put a working product in front of customers for a fraction of what it cost a decade ago.
The trade-offs deserve naming rather than glossing over. Open-source components carry maintenance and security obligations you inherit along with the code. Renting infrastructure and models means depending on suppliers who can change prices or terms. Startups that treat these as free tend to discover the costs later. Our guide to building an open-source strategy covers how to use community-built software without being caught out by it, and cybersecurity architecture for distributed businesses deals with the security half.
The wider effect is that fewer companies now fail for lack of technical capability. They fail for lack of customers, which is a different problem and rarely solved by better engineering. That is why product-led growth and community-driven distribution have become standard playbooks rather than novel tactics.
What to Watch Over the Next Two Years
Predictions about the next decade are mostly entertainment. Two years out, a few things look reasonably clear.
Concentration will either normalize or become structural. If a small number of AI labs keep absorbing most available capital, the market for everyone else stays tight regardless of headline totals. Watch the share of funding going to rounds under $50 million rather than the overall figure.
Exit activity will set the tone for seed funding. Investors who get money back deploy it again. A second strong year for IPOs and acquisitions would loosen the early-stage market more than any change in sentiment.
Vertical specialization will keep beating general-purpose tools. Software built for one industry, with its rules and vocabulary already encoded, is harder to displace than a flexible tool that fits everyone approximately. This is why small, focused software companies continue to find defensible niches.
Hiring will stay flexible. Small teams supplemented by contractors have become the default rather than a cost-cutting measure, which our coverage of distributed team productivity examines in more detail.
Conclusion
Startups shape business trends less by inventing categories than by resetting expectations, and the 2026 evidence supports a measured reading rather than a dramatic one. Business formation is high but mostly small-scale. Funding is at a record level and unusually concentrated. Failure is common but less common than the folklore claims. AI is everywhere and therefore no longer an advantage on its own.
For anyone running a business, the practical takeaway is about standards rather than technology. The things startups made normal, fast onboarding, transparent pricing, support that answers, personalized service, are now what your customers expect from you too. Meeting that standard matters more than tracking which sector is currently attracting capital. Our overview of business model innovation and the guide to scaling a young company go further into how to act on that.
Found this useful?
Make SmartKeys a preferred source on Google, and our articles will surface more often in your Top Stories, AI Overviews, and AI Mode.
Add as Preferred Source
