The most quoted number in venture this year is that AI absorbed 86% of US venture spending in the first half of 2026. It is true, and close to useless if you are raising a seed round, because it measures dollars, and dollars are dominated by a handful of rounds that have nothing to do with your market. Change the lens, and the picture changes completely.

Three measures of AI funding in H1 2026:
86% of US VC dollars · 43.2% of US VC deals: PitchBook and NVCA
49% of pre-seed dollars: Carta

Eighty-six becomes forty-three the moment you count deals instead of dollars. Even those figures describe different datasets from Carta's pre-seed sample; the point is to choose the measure that fits your round.

Fintech has a different AI mix too. F-Prime's 2026 State of Fintech report shows AI companies taking 37% of fintech funding versus 72% of software funding in Q4 2025. Those are global figures for an earlier period, not two more H1 2026 US statistics. SVB's 2025 fintech report also found a lower AI share in fintech: 30%, against 58% across venture in its analysis. There is room for AI in fintech without assuming that every financial business must sell an AI story.

Investor perspective: Anish Acharya, a16z (27 April 2026). His thesis is that AI agents could remove the friction that keeps consumers in expensive or unsuitable financial products. That is a specific fintech opportunity, rather than a funding argument built around the AI label. Read the post on X.

One caveat. The US has the deepest stage-level dataset, so the next three sections are American unless stated otherwise.

Carta's Q2 2026 data is the clearest read on the bottom of the market: $3.19bn across more than 11,500 instruments, against $3.22bn across 14,825 a year earlier. The same money into roughly 22% fewer instruments. The average instrument hit $276,000, up 27% and a record.

Round size tells a second story. Carta separates rounds below and above $1m; compare the lighter H1 segments, since 2026 is only a half-year observation.

The mechanics have standardised. In Q2, 93% of pre-seed rounds were SAFEs and 91% of SAFEs were post-money. Across H1, 94% of post-money SAFEs had valuation caps. The Q2 median cap on SAFEs above $2.5m reached $35m, up 40%.

Specialist capital: Sheel Mohnot, Better Tomorrow Ventures (1 October 2025). BTV raised a $140m fund to back fintech founders at pre-seed and seed. It is a concrete example of a stage-specific mandate; this is fund capital announced in 2025, not another H1 2026 startup-funding statistic. Read the post on X.

This is the finding that surprised me most, and it comes with a sample-size warning I want stated before the numbers. Using Carta's round benchmarking tool, which covers roughly 20,000 priced US rounds since 2021, fintech seed has gone from trading above the market to trading below it.

Fintech seed valuations went sideways while the wider market re-rated 17%, and fintech seed round sizes fell 9.5%. Meanwhile fintech Series B is running hot: $240m median pre-money against $145m for everything else, at 10% dilution.

Treat that as a signal rather than settled fact. The H1 fintech cells rest on small samples, 39 seed rounds and 22 Series B, and Carta's counts revise upward as customers file. Carta also sees only its own cap tables: PitchBook counted 5,674 first-time financings in H1, on pace for a record year, which cuts the other way.

Dispersion inside the stage is extreme. Peter Walker at Carta put the 95th-percentile seed valuation at $200.4m in Q2 against $72.2m a year earlier, a 177% jump versus 92% at the peak of the 2021 frenzy. Against a median under $19m, that is an eleven-fold spread inside one round name.

Data perspective: Peter Walker, Carta (8 July 2026). Read the seed-valuation post on X.

SVB's 2025 Future of Fintech report has the most useful statistic for any fintech founder planning a raise: companies closing a Series A in the 24 months covered by that report had $4m in median annual revenue, up from $1m four years earlier. It is a fintech-specific annual revenue benchmark at the time of the raise. Use the same revenue definition when comparing your own business; an annualised run rate is not automatically equivalent.

Context from SVB's State of the Markets, H1 2026, all sectors, median revenue at time of raise. 2021 → 2025, all sectors:
Seed: $156k → $363k
Series A: $1.6m → $3.3m
Series B: $5.8m → $7.1m
Series C: $14.5m → $18.9m

The all-tech bar doubled. The fintech bar quadrupled, and now sits about 21% above it. These are historical medians from different samples, not universal minimums for a Series A. SVB puts the seed-to-A jump at 8x to 12x revenue growth, and notes the middle 50% raise on between $1m and $6.5m.

Dilution moved in founders' favour at the same time. Carta's benchmarks put median seed and Series A dilution at 18% and Series B at 12%, with founders retaining 56% after seed and 36% after Series A.

Carta publishes benchmark bands rather than observed rates, with a US mid-case of 10% reaching Series A after one year, 25% after two and 35% after three. For a historical European comparison, Dealroom tracked 3,075 companies in the 2016 to 2018 seed cohort: 6% raised a Series A within 12 months, rising to 27% by 36 months. Latin America is starker: LAVCA found two-year graduation collapsed from 16% for the 2019 cohort to 3% for the 2022 cohort.

These are different cohorts and methodologies, so they are not a current league table of regional odds. The planning lesson survives: allow at least two years between rounds, and stress-test a longer gap against the conditions in your own market.

The runway test: What happens if the next round takes 12 months longer than your base case? Build that answer before you set the burn rate.

SVB’s 2025 report shows how founders were already adapting: median net cash burn fell 12% year on year in Q2 2025, the eighth consecutive quarter of reductions. A longer funding gap needs a lower burn rate.

The global story is not that every region follows the US at a lower valuation. Each has a different constraint.

The headline global number hides all of it. KPMG recorded $103.1bn of global fintech investment in H1, putting 2026 on pace for its strongest annual result in four years across VC, PE and M&A combined, while deal volume fell from 2,501 in H2 2025 to 2,100. EMEA managed $11.3bn across 626 deals, on pace for a decade low. Asia Pacific fell to $4.6bn, though India went the other way at $2bn across 101 deals. China recorded just $149m, and KPMG is explicit that this understates reality: fintech there has been "internalized by banks, insurers and large platforms," outside conventional VC data.

KPMG’s largest-deals map makes the distinction visible: acquisitions and buyouts sit alongside venture rounds. A global investment total is not the same thing as capital available to seed founders.

Africa is the sharpest counterexample. On Partech's 2025 numbers, African fintech raised $769m of equity, 32% of all African tech equity, plus $716m of debt, 44% of the continent's debt funding, with Seed+ rounds about 35% larger than other sectors. That is a round-size premium, not a valuation premium, so it cannot be compared directly with the US seed pricing table. It is still a clear sign that the sector carries a different weight in Africa. There is no single global fintech funding cycle.

In Latin America, fintech took 61% of all venture capital on LAVCA's 2024 data from 30.6% of deals. The case there is financial inclusion and payment rails, not AI saturation.

None of this means the concentration is imaginary:

Even at the height of the enterprise software boom in 2001, only 38% of North American venture went to B2B software. The squeeze is real. It just operates at the top of the market, through vehicles that were never going to fund a seed round.

Watch: Peter Walker on the maths behind the raise. In this May 2026 Post Money conversation with Nilanjana Bhowmik, Carta’s head of insights discusses dilution, funding concentration, and the long path to an exit.

NestiFi has raised from Argonautic Ventures. I include it because of what Argonautic is: a firm that describes itself as having AI at its core, writes first institutional cheques at pre-seed and seed, and names financial technology as its first vertical. That is not a fintech fund making an AI exception. It is an AI fund for which fintech is a primary thesis, operating at exactly the stage where the 86% statistic stops applying. This follows on from our recent successful raise on Spark Venture Funding. For more on our own fundraising experience: We're Raising in Public. It's Buying More Than Capital.

Finch Capital found that every European fintech round above €1bn between 2021 and 2025 was US-led, while the sub-€100m early-stage segment is largely self-sufficient on European capital. Europe’s dependence on US-led capital rises sharply at the largest round sizes.

Irish fintech had a strong 2025 on Finch's numbers, at €228m across 26 deals, up 124%. But TechIreland names the constraint: a record 211 Irish companies raised less than €1m in 2025, while only 25 raised €3m to €10m against a historic average of 30 to 40. Ireland's problem is not seed formation. It is the step after it.

Two points in Europe's favour. Atomico's data shows the AI valuation premium barely exists at the bottom of the market, at about 19% above non-AI at seed and 22% at Series A, widening to 161% by Series C. European seed pricing is not distorted by the AI trade the way American pricing is. And fintech takes 9.2% of European venture funding against 3.1% in the US.

Investor perspective: Augmentum Fintech / Tim Levene (26 January 2026). Augmentum emphasises execution, founder quality, and market readiness in early-stage investing. Benchmarks frame the conversation; they cannot replace evidence that a particular team can build the business. Read the post on X.

Pre-seed. $276k average instrument; SAFEs dominate US pre-seed. Model a capped post-money SAFE and the combined dilution of stacked instruments.

Seed. $18.1m fintech median pre-money; $200m at the wider seed market’s 95th percentile. Plan around the median. Model roughly 20% dilution and at least 24 months of runway.

Series A. $4m median annual fintech revenue in SVB’s 2025 report. Treat it as a benchmark, not a cutoff. If well below it, keep a seed-extension option open.

Series B. $240m fintech median pre-money at 10% dilution, in a small sample. Test investor appetite early, while checking comparable companies and terms.

Two market-wide facts. Carta has the down-round rate at 11.4%, down from a 22% peak in 2023: selective, not distressed. And in an analysis of descriptions of 9,000 venture-backed companies, SVB found 42% of those marketing themselves as AI show little evidence AI is central. The label is now noise.

Non-AI venture funding has roughly halved since 2021, and I am not going to pretend otherwise. But the 86% figure describes a market for foundation models, not one for financial infrastructure, and it describes one country.

Fintech seed and pre-seed drew $2.6bn globally in H1, and early-stage fintech deal value is running ahead of every year except the 2021 and 2022 peaks. QED and McKinsey describe the shape as a barbell, with midstage companies squeezed and both ends funded.

Investor perspective: QED Investors (19 June 2026). QED sets out the barbell thesis directly: capital at both ends, with a tougher path through growth equity. Read the post on X.

What differs is where the friction sits. Selection in the US, follow-on in Europe, measurement in China, capital structure in Africa, graduation in Latin America. Work out which constraint is yours, then raise against that rather than against a headline about somebody else's market.

Explore the numbers: Open Carta’s round benchmarking tool and filter by stage, sector, and period.

If you found this analysis valuable, subscribe to Synaptic Finance and connect with me on LinkedIn for weekly deep dives at the intersection of AI, blockchain, and the future of finance.