Startup Statistics 2026: 52 Numbers From 235,039 Launches

Ten numbers, one table. Every one is unpacked and sourced further down the page.

235,039 dated business-idea launches were collected between August 2025 and June 2026 in the Fluenta index, across 146 launch, funding, and discussion sources. Nobody has assembled a bigger public record of what actually got shipped, pitched, and argued about.

Product Hunt nearly tripled in ten months. About 7,300 launches in August 2025. About 21,600 in June 2026. It is the one source with an unbroken monthly record, and the step change landed in March.

The wave is rotating. From January to June 2026, Productivity and Collaboration gained 4.1 points of launch share and Marketing and Creator gained 2.6, while Developer Tools lost 3.8 and generic AI and Automation lost 3.4. Tools for builders are ceding the launch wave to tools for work. Quietly, but in every month of data.

A note on method first: these failure numbers come from scoring, not surveys. Each idea gets graded on six live signals: demand, pain, urgency, competition, monetization, funding. Score under 32? At least two signals failed outright.

90.7% of ideas die before scoring even begins. Of 4,238 idea mentions scouted from 74 active sources in May 2026, only 393 were distinct and serious enough to score.

Urgency failed in 12 of 12 killed ideas. Demand failed in 11. The two cheapest signals to check do nearly all the killing.

Monetization killed just one. Funding appetite failed in 9 of 12, pain in 4, competition in 3. Almost every dead idea could have made money, if anyone had needed it.

Six weak signals in one week has no survivors. Six independent signals reading weak in the same week is the cleanest failure-guaranteed pattern in the dataset.

Public sources tell the same story one stage later. Per US Bureau of Labor Statistics survival data, roughly one in five US businesses closes in its first year and about half are gone within five. Per CB Insights, the most cited reason startups fail is building something with no market need. Same disease, caught earlier here.

Score your idea on the same six signals in 20 minutes →

130 SaaS ideas, scored in Q2 2026 on two axes: can this get funded, and is the market already full. Four quadrants fall out, and the good one is nearly empty.

The same picture as a table:

The median idea walks into a market that is already 57% saturated. Before a line of code gets written.

Too late has a number, and the number is 70. Saturation above 70% marked every idea in the too-late group. And no competition is not the green light it looks like: 19 ideas with zero competitors turned out to have zero market.

Only 31 of 89 AI-native ideas are defensible for a new entrant. That is 34.8%. The other 58 split between ghost markets, where demand evaporates the moment anyone checks, and categories already too crowded to enter.

The AI agents market is projected to grow 10x by 2030, from $5.26B in 2024 to $52.62B, a 46.3% compound annual rate. Vertical AI agents grow fastest at a projected 62.7% CAGR, well ahead of general-purpose agents.

Open weights move in days; closed models leave a gap. Gemma and Qwen spawn derivative launches on GitHub and Hugging Face within 0 to 3 days. Claude and GPT reach launch platforms in 1 to 6 days and leave Hugging Face empty.

A model release barely moves the market. A Claude release lifts launch activity by 1.07x on average. Sustained mindshare beats the release-day bump.

AI and Automation reads as over-hyped in the sector map: huge launch share, merely average quality. May showed the same split. June showed it again. That is a structure, not a blip.

Check whether AI engines recommend your business, free →

All 194 companies in the YC Spring 2026 batch, scored on six public signals before Demo Day. The company-by-company table lives in 194 YC Spring 2026 Startups, Scored on Public Data.

Four demo-day groups, all of them AI. The AI Workforce with 62 companies, Agent Infrastructure with 49, Care and Capital with 44, AI Meets the Real World with 39. There is no non-AI YC anymore. The most common category: AI and Automation with 53 companies, then Developer Tools with 22, Healthcare with 18, Robotics with 17.

The averages sit in a four-point band. Agent Infrastructure 50.3, AI Workforce 49.1, Care and Capital 48.3, AI Meets the Real World 46.3. Batch average: about 48.7 of 100.

14 of 49 priced Agent Infrastructure companies need over a year to earn back one customer. Strong monetization scores, broken payback math.

18 of 19 health companies carry an FDA or reimbursement barrier that no public signal measures. The score cannot see the regulatory wall. The wall is still there. Meanwhile the Care and Capital monetization pillar averages 89%, because health and money businesses take a slice of transactions that already happen.

Scouts versus stats, the oldest argument in baseball, replayed with business ideas. The trend tools are the scouts. Here is what the stats said.

47 trending ideas averaged 47 of 100 on the same rubric as the YC batch. Scores ran from 31.3 to 65.5, and exactly one idea cleared 60. Trending and buildable turn out to be different lists.

The typical trend-spotted idea draws 275 searches a month. Genuine. Early. Tiny.

The one that cleared the bar was a plain SEO audit platform. It scored 65.5, matching the strongest company in YC's entire Spring 2026 batch. The most buildable trending idea of the summer was a boring service with paying customers attached.

Search demand is the signal that separates winners. Across 1,238 ideas scored from March to June 2026, it split the top fifth from the bottom fifth by about 28 points, in every single month. No other signal came close. The Business Idea Demand Index, which compresses quality, demand, monetizability, and novelty into one monthly reading, rose from 50.4 in March to 54.3 in June.

The top three ideas of H1 2026: Python automation scripting at 66.2, an SEO content workflow engine at 66.2, an AI SaaS coding flow at 65.9. Boring, monetizable work beat frontier hype across the entire half. The ten highest, as a preview; the full top 20 with six-signal breakdowns lives in the H1 report:

Run your idea through Fluenta before you build it →

Fluenta scores business ideas on six live market signals: search demand, audience pain, urgency, competition density, monetization, and funding appetite. This page aggregates 235,039 launches collected across 146 sources between August 2025 and June 2026. Within that record, 1,238 ideas were fully scored between March and June 2026. Cohort analyses cover 130 SaaS ideas from Q2 2026, 89 AI-native ideas, 47 trend-spotted ideas, 393 ideas scouted in May, and all 194 YC Spring 2026 companies. Scores use public data only, which cuts both ways: a company with quiet public signals and signed private pilots can score low. The scores measure market evidence, not team quality.

Source: Fluenta, Startup Statistics 2026. fluenta.space/resources/guides/startup-statistics

Every chart on this page is free to republish with a link to this page as attribution. The reports behind each section carry the full datasets: the H1 2026 record, the saturation report, the AI ranking, the YC batch scores, the kill list, and the trend tracker.

Changelog: July 2026, page created from the H1 2026 dataset.