Global venture funding hit a record $510 billion in the first half of 2026, more than all of 2025, yet far fewer startups are getting funded: a handful of AI giants took the lion's share while seed deal counts and India's deal volume fell sharply.
The record: $510 billion in six months
Crunchbase data shows the first half of 2026 beat the whole of 2025 ($440 billion), and the money was heavily concentrated.
Period | Global venture funding | What stood out |
|---|---|---|
Q1 2026 | $305 billion | Largest quarter Crunchbase has tracked; initially reported near $297–300 billion, later revised up |
Q2 2026 | $205 billion | Second-largest quarter ever; 5,000+ startups funded |
H1 2026 | $510 billion | Above all of 2025 ($440 billion) |
The concentration is the real story:
OpenAI and Anthropic alone raised $217 billion, or 43% of all H1 funding.
Anthropic's $65 billion round was close to a third of Q2 funding.
Sixteen billion-dollar rounds totalled $108.6 billion, or 53% of Q2.
More than 70% of Q2 capital went to AI companies, and about two-thirds to US startups.
Exits reopened too. Q2 set records for both IPOs and acquisitions of venture-backed companies, including the largest venture-backed IPO and the largest startup acquisition ever, both involving SpaceX. Thirty-two companies went public above $1 billion in the quarter.
The barbell: giant rounds at one end, thin seed at the other
Late-stage funding reached $134 billion in Q2, up 141% from a year earlier, while global seed funding was $12 billion. Seed's share of all venture dollars fell from 11.3% to 5.9% over the same period, according to Axis Intelligence's calculation on Crunchbase stage totals.
In Q1, seed dollars rose 31% to $12 billion, yet seed deal count fell about 30% to roughly 3,700–3,800. In plain terms: bigger cheques to fewer founders. Part of that is a handful of AI "seed" rounds that run into the hundreds of millions, which inflate the totals while the traditional seed band shrinks.
For a founder, the headline number is therefore misleading. Record funding does not mean it is easier to raise; it means capital is pooling at the top. Non-AI startups feel it most, and several analysts report that AI companies command a seed valuation premium of around 42% over non-AI peers.
India: more dollars, far fewer deals
Tracxn's India Tech report for 1 January to 21 September 2026 shows the same barbell on a smaller scale.
Metric (9M 2026) | Value | Change vs 9M 2025 |
|---|---|---|
Total funding | $10.3 billion | +7% |
Funding rounds | 1,134 | −38% |
Seed funding | $698 million | −37% |
Early-stage funding | $4.2 billion | +27% |
Late-stage funding | $5.4 billion | roughly flat |
New unicorns | 6 | up from 4 |
AI infrastructure was the most-funded theme at $1.2 billion, ahead of digital lending ($799 million) and payments ($773 million). Bengaluru took 43% of all tech capital ($4.4 billion). Tracxn notes the new unicorns reached scale on markedly less capital than earlier generations.
Weekly data shows how top-heavy it is: of the $203.4 million Indian startups raised from 21 to 25 September, Ultraviolette ($85 million) and Ema ($77 million) made up nearly 80%, per Inc42.
Note on data: Inc42 counted $5.2 billion for H1 2026 (down 9%) with deal count up 7%, a different scope from Tracxn's tech-only, nine-month view. Treat trackers as directional and compare like with like.
What founders should do now
These five moves are Startup360's read of the data, not guarantees:
Raise on proof, not a story. Investors are backing paying customers, signed LOIs and defensible data or IP. Show active paying users, not total sign-ups.
Be specific about where you sit. Investors now ask whether you build models or use them. Vertical focus (health, legal, finance, construction) plus a clear answer to "what if the foundation models absorb this?" is the more fundable pitch.
Don't copy frontier-lab capital needs. Most early-stage companies can reach revenue on far less than a mega-round, and several analysts see revenue-first and seed-strapping paths gaining ground.
Avoid AI-washing. Relabelling a non-AI product as AI invites scrutiny; build AI into the core product or pitch the business on its own merits.
Plan longer runways. With fewer seed cheques and a thin middle, 18 months of cash is a minimum, not a luxury.
What to watch next: Crunchbase's Q3 2026 numbers will show whether the concentration eased or deepened, and TechCrunch Disrupt runs 13–15 October in San Francisco, a likely venue for fresh funding signals.
