The State of Web3 Startups 2026

Capital, Institutional Adoption, and the New AI Discovery Layer

Published by Generis, a GTM-first agency working with digital asset platforms, Web3 infrastructure, and businesses building in Web3.

Generis is an official partner of the European Blockchain Convention 2026 (Barcelona, September 16-17, 2026).

About This Report

This report examines how three developments intersect for Web3 companies: venture capital becoming highly concentrated, regulated institutional participation in crypto being visible in public registers, and AI becoming a routine part of B2B research and vendor evaluation.

The evidence base is limited to published venture and market data, regulatory and tokenization datasets, and buyer behavior and web traffic research from named research producers. The scopes of the figures coming from different datasets are kept separate. Derived figures are labeled as calculations.

Every quantitative claim is linked to a publicly verifiable source. Claims that could not be reproduced from a primary source, a directly attributable research producer, or a clearly dated public snapshot have been removed from this edition.

Executive Summary

1. European blockchain funding is a small and concentrated part of the global market.

CV VC reports that Crypto Valley raised $728 million across 31 deals in 2025, which is 47% of European blockchain funding and 5% of global blockchain funding in its dataset. Those figures imply that Europe accounted for roughly 10% of CV VC’s global blockchain funding universe. Separately, Galaxy Research found that 11 deals accounted for 85% of the $8.5 billion deployed globally in Q4 of 2025.

2. Europe’s regulated crypto layer is now substantial and visible in public data.

MiCA came into effect on 30 December 2024. The maximum transitional period under Article 143 ended on 1 July 2026. Outrun Advisory’s 31 July 2026 snapshot of the ESMA register counted 336 MiCA-authorized CASPs across the EEA. Its directory includes crypto-native firms as well as banks, investment firms, and payment institutions. Outrun’s business-type classification is an analytical overlay.

3. AI has become part of B2B discovery and vendor selection.

Forrester reports that 94% of business buyers use AI in their buying process, based on research covering more than 17,500 global business buyers. In a separate G2 survey of 1,076 B2B software buyers, 51% said they begin software research in an AI chatbot more often than in Google, and 69% ultimately chose a different vendor than originally planned.

4. Discovery and trust are not the same problem.

Co-branded SurveyMonkey and Reddit research covering 1,202 US business decision-makers found that 73% trust peer recommendations during purchasing, compared with 39% who trust AI chatbots. AI can help shape consideration, but independent credibility remains a separate requirement.

Part 1: The Capital Environment

1.1 Concentration instead of contraction

The headline totals for crypto venture capital look strong. The distribution is highly concentrated.

Galaxy Research reported $20 billion invested across 1,660 crypto deals in 2025, which is the largest annual total since 2022 and more than double the 2023 total.

All figures below come from the same Galaxy report:

Metric	Value	Period
Share of capital allocated to later-stage companies	57%, the largest share observed in Galaxy’s dataset	Full year 2025
Deals above $100M as a share of quarterly total	11 deals = 85% of the $8.5B quarterly total, about $7.3B combined	Q4 2025
Median deal size	$4M	Q4 2025
Median pre-money valuation	$70M	Q4 2025

Galaxy also argues that the industry is maturing and that the earlier era of broad pre-seed crypto venture investing may be ending as established companies move into categories that previously required startups.

Q1 2026 recorded approximately $4 billion across 355 deals, a 50% quarter-on-quarter decline in dollars and 16% in deal count. Median deal size reached an all-time high above $4.5 million, and the pre-seed share of deal count fell to 19%. Galaxy notes that annualizing the Q1 run rate would produce roughly $16 billion for 2026. This is an illustrative annualization.

1.2 Europe’s structural position

CV VC’s 2025 dataset provides a directly sourced view of European concentration.

The 11th Crypto Valley Top 50 & Ecosystem Report release states:

Crypto Valley raised $728 million across 31 deals in 2025, up 37% year-on-year.
That total represented 47% of European blockchain funding and 5% of global blockchain funding in CV VC’s dataset.
Crypto Valley hosted 1,766 active blockchain companies, up 134% since 2020.
Zug accounted for 41% of active companies, 20 of the 31 deals, and 88% of disclosed capital.
The combined valuation of the Crypto Valley Top 50 reached $467 billion, including ten unicorns.

Those primary figures allow two useful calculations. If $728 million represents 47% of European blockchain funding, the implied total for Europe is approximately $1.55 billion. Against CV VC’s reported global blockchain funding total of $15.5 billion, Europe therefore represents approximately 10% of that dataset.

Concentration inside Europe is also visible in the same source. CV VC lists TON as the largest Crypto Valley deal of 2025 at $400 million. That single transaction equals about 55% of Crypto Valley’s $728 million annual total and approximately 26% of the implied European total.

Dataset note: CV, VC, and Galaxy use different deal universes and report different global totals for 2025. Their figures should not be combined as though they were a single dataset. The concentration patterns are discussed separately above.

1.3 Speculative token failure is widespread in the GeckoTerminal dataset

CoinGecko analyzed cryptocurrency listings on GeckoTerminal from July 2021 to December 2025. Its methodology counts assets that had at least one completed trade and later ceased active trading. For pump.fun, only tokens that graduated to a decentralized exchange are included.

The findings from CoinGecko Research are stark:

53.2% of cryptocurrencies in the studied GeckoTerminal dataset had ceased active trading by the end of 2025.
11,564,909 ceased active trading during 2025 alone (86.3% of all failures in the dataset).
Annual failures were 2,584 in 2021, 213,075 in 2022, 245,049 in 2023, 1,382,010 in 2024, and 11,564,909 in 2025.

The 2025 surge was largely attributed by CoinGecko to low-effort token creation through platforms that reduced the technical barrier to launching new assets.

Scope matters. Adding the annual figures produces approximately 13.4 million failures across the July 2021-December 2025 observation window. This is a count within CoinGecko’s defined GeckoTerminal universe, not a claim about every token ever launched globally.

Part 2: What Buyers and Investors Now Require

2.1 Commercial evidence is appearing at an early stage

The State of Web3 Capital 2026, produced by INPUT Global in collaboration with Proof of Talk, combines more than 200 Proof of Pitch applications submitted between January and May 2026 with a survey of 13 Web3 fund partners and ecosystem leaders.

On the founder side:

44% of applicants had already generated revenue, and 7% reported profitability, while 89% were still raising at the pre-seed or seed stage.
RWA and tokenization were the primary focus for 29% of founders, ahead of DeFi at 23%.
Only 5% sought token-only financing, 83% wanted some form of equity exposure.

On the investor side, 92% of the 13 respondents selected RWA and tokenization as a priority area, with DeFi and stablecoins tied at 77%.

Scope note: The investor sample contains only 13 respondents and is described by the publisher as fund partners and ecosystem leaders, not 13 statistically representative venture funds. The applicant pool is also self-selected through a single-pitch program. These findings indicate direction within that sample, not market-wide weighting.

The publisher’s conclusion is that founder focus and investor priorities are showing greater alignment than in the previous cycle.

2.2 The institutional layer in Europe

MiCA became fully applicable on 30 December 2024. Under Article 143, Member States could allow existing providers to continue under transitional arrangements for up to eighteen months. The maximum period ended on 1 July 2026, while individual states could apply shorter periods or none at all.

Outrun Advisory’s MiCA CASP dashboard, based on the ESMA public register, counted 336 authorized CASPs across the EEA in its 31 July 2026 snapshot.

The same directory includes crypto-native exchanges, brokers, custodians, and infrastructure providers, as well as banks, investment firms, and payment institutions. Outrun classifies those business types independently. MiCA and ESMA do not maintain an official crypto-native-versus-TradFi category.

This distinction matters. The existence and authorization status of CASPs come from the regulatory register. The business-model labels are a third-party analytical layer. The reliable conclusion is therefore narrower than a market-share claim: traditional financial institutions are already present in the MiCA-authorized provider population.

Dynamic-data note: both the ESMA register and dashboards built on it change as authorizations, amendments, and withdrawals are reported. Any count should be read as a dated snapshot rather than a permanent total.

2.3 The tokenization numbers, read correctly

Tokenized real-world asset datasets can produce very different headline totals because trackers distinguish between different forms of tokenization.

RWA.xyz separates distributed assets, which can move peer-to-peer on public rails, from represented assets, which are recorded on a ledger but cannot leave the issuer’s platform. On 12 March 2026, RWA.xyz reported $26.55 billion in distributed value and $342.60 billion in represented value, excluding stablecoins. Those figures describe different categories and should not be substituted for one another.

A separate BeInCrypto Research study, using data as of 31 May 2026, tracked roughly $60 billion across more than 7,000 products. Its analysis found that 97% of the tokenized asset value in its dataset was outside US retail reach, leaving approximately $1.7 billion, or about 3%, retail-accessible.

Scope note: the 97% figure is BeInCrypto Research’s access classification for its dataset. It is not a universal on-chain statistic and should be cited accordingly.

Part 3: AI Has Become Part of The Discovery Layer

3.1 Many buyers now use AI during research

Forrester’s Buyers’ Journey Survey, 2025 (the dataset underpinning its 2026 Buyer Insights research and covering more than 17,500 global business buyers) found that 94% of business buyers report using AI in their buying process. In related Forrester reporting, 55% used AI for product comparisons, 54% for product information or research, and 47% for building a business case.

A separate G2 study of 1,076 B2B software buyers, surveyed in March 2026 across North America, EMEA, and APAC, found:

Finding	Figure	Change reported by G2
Use AI chatbots for software research	71%	up from roughly 60% seven months earlier
Begin software research in AI more often than in Google	51%	–
Choose a different vendor than originally planned	69%	–
Choose a vendor not previously considered	33%	–
Think more highly of a vendor cited by an AI	85%	–

These are two distinct studies with distinct populations. Forrester surveys business buyers across categories. G2 surveys B2B software buyers. Their percentages should not be combined into a single funnel or treated as the same denominator.

Taken together, the studies establish that AI is being used during research and can affect vendor consideration. They do not prove that every B2B shortlist begins in AI or that the same behavior occurs at the same rate in every Web3 buying process.

3.2 AI referral traffic is small in volume but can show high-intent behavior

Conductor’s 2026 benchmark found that AI referrals accounted for 1.08% of website traffic across ten major industries. The traffic analysis covered 1,215 enterprise customer domains and more than 3.3 billion sessions between May and September 2025. Conductor used a separate 13,770-domain dataset for AI citation and visibility analysis.

Conversion studies point in the same general direction but measure different events and different markets:

Producer	Finding	Sample and scope
Semrush	Semrush estimates that an LLM-referred visitor can be worth about 4.4× an organic-search visitor, based on conversion-rate relationships	500+ high-value digital-marketing and SEO topics, modeled the visitor-value relationship, not a universal observed conversion rate
Opollo	14.2% AI-referral conversion vs 2.8% Google organic	312 IT and technology service firms in North America, Australia, and the UK. Data collected January 2025-January 2026
Microsoft Clarity	1.66% LLM-referred signup rate vs 0.15% from search	1,277 publisher and news domains, one month of data
Adobe Analytics	AI-referred US retail traffic converted 42% better than non-AI traffic in March 2026, after a 38% deficit in March 2025	More than one trillion visits to US retail sites, retail purchase conversion, not B2B lead conversion

The Adobe series is notable because the relative relationship reversed within twelve months in the same research program. But the studies above should not be pooled into one benchmark.

Conversion is not defined uniformly across these studies: signups, purchases, and qualified inquiries are different events. Opollo also excluded firms with fewer than ten monthly AI sessions and conversion outliers above 35%. The evidence supports a high-intent hypothesis in several verticals, it does not establish a universal AI-referral conversion rate.

3.3 AI is not a substitute for trust

Co-branded SurveyMonkey and Reddit research covering 1,202 US business decision-makers found that 73% trust peer recommendations during purchasing, compared with 39% who trust AI chatbots.

That result places an important boundary around the AI-discovery narrative. AI can be part of research and consideration, while peer and third-party validation remain separate trust mechanisms.

For institutional Web3 go-to-market, the evidence supports treating discoverability and credibility as distinct problems: being findable can help a company enter the consideration set, while independent coverage, references, and verifiable evidence address the trust layer.

Part 4: What the Evidence Supports

Three conclusions can be made without relying on proprietary client audits or unverifiable historical snapshots.

European blockchain capital is concentrated. CV VC’s own 2025 figures indicate that Europe accounted for approximately 10% of its global blockchain funding universe, while Crypto Valley captured 47% of the European total. One $400 million transaction represented roughly a quarter of the implied European total. Galaxy’s separate dataset shows a similar global concentration pattern in Q4 2025, with 11 deals accounting for 85% of deployed capital. The datasets differ, but both show that aggregate funding totals can conceal extreme concentration.
Institutional participation is already visible in the regulatory layer. MiCA is fully applicable, the maximum transitional period has ended, and the public CASP register contains hundreds of authorized firms. Outrun’s directory shows that the population includes both traditional financial institutions and crypto-native providers. That does not quantify institutional demand for every Web3 product, but it does establish that regulated institutional participants are already inside the European crypto market structure.
AI is already material to B2B research and vendor consideration. Forrester finds widespread use of AI across the business buying process. In contrast, G2 finds that a majority of surveyed software buyers begin research in AI more often than in Google and that AI-assisted research can change vendor choice. These surveys do not prove causality for European Web3 deals specifically, but they establish a new discovery environment that Web3 companies selling to institutions cannot reasonably ignore.
The practical implication is narrower than a promise of a pipeline. Technical discoverability, clear category positioning, evidence from founders and the company, and credible third-party references are increasingly relevant risk controls for institutional go-to-market. The evidence in this report supports investing in those assets. It does not establish a universal conversion uplift or guarantee that visibility alone produces revenue.