812 YC Fintech Companies, an 88% pipeline drop, and the Ideas Worth Rebuilding
YC’s largest vertical has a 36% lower closure rate than average — but the 91 that failed left behind a blueprint for AI-era financial infrastructure.
One in seven YC companies touches money.
812 companies — 14.2% of all YC — carry a fintech tag. Add keyword expansion and the number reaches 862. It is the single largest vertical in Y Combinator’s portfolio, ahead of healthcare, developer tools, and B2B SaaS.
Fintech vs YC Overall
In 2022, Y Combinator funded 165 fintech startups. In 2026, it funded 19.
The ZIRP era inflated fintech batches to unsustainable levels. Zero interest rates made every financial product a venture bet — neobanks, BNPL, crypto exchanges, embedded finance. Then rates rose, FTX collapsed, and regulators tightened. The pipeline contracted 88% in four years.
The peak wasn’t just a batch-size effect. Fintech’s share of each YC batch surged from ~10% in 2015 to over 20% in 2021-2022, then cratered. YC didn’t just fund fewer companies — it lost faith in the category.
But look at the closure rates. Despite the boom, fintech’s 11.2% closure rate is 36% below the YC average. The companies that got in were stronger — or the product (a transaction) hits revenue faster than most.
Crypto startups show six times the closure rate of fintech infrastructure.
Crypto & Web3 has a 17.6% closure rate — actually above the YC average. Fintech Infrastructure sits at 2.9%. The spread is the widest of any vertical we’ve analyzed. Crypto behaves more like speculative tech than financial services. Infrastructure holds up because it sells picks and shovels regardless of which consumer bets pan out.
Closure Rate by Sub-vertical
Balanced built marketplace payments in 2011. FedNow launched in 2023.
The most striking pattern in inactive fintech companies: the thesis was right, the timing was wrong. Regulatory changes, infrastructure shifts, and AI breakthroughs arrived years after these companies burned through their runway. The gap between vision and viability was measured in regulatory cycles, not product cycles.
Marketplace payments API
FedNow + agent commerce
Agent-to-agent transactions reopen marketplace payments without card networks
Social micropayments
Stablecoin L2 rails
Sub-cent transactions viable on USDC/L2 for agent economy
Plaid for SE Asia
Open banking mandates live
Indonesia, Thailand, Philippines all launched open banking frameworks
Cannabis banking
Schedule III + SAFE Banking
$35B market, still no banking. Regulatory blocker lifting
App for unbanked workers
FedNow + employer channel
Instant settlement + earned wage access as acquisition channel
Programmatic insurance
$700B embedded insurance
AI underwriting + platform demand + climate urgency
Gen-Z financial advisors
LLM advisors + aged TAM
Gen-Z is 22-28, earning. LLMs replace $200/hr human advisors
NLP algorithmic trading
LLMs solve NLP-to-logic
Claude/GPT-4 parse financial strategies from natural language
We scored all 91 inactive fintech companies on rebuild viability.
The scoring considers changes in market size, infrastructure, and operating costs since each company closed. Ten companies scored 7 or above, suggesting their ideas may be worth another look.
A payments platform for two-sided marketplaces — escrow, split payments, seller onboarding, and payouts through a single API. Stripe Connect before Stripe Connect existed.
Stripe launched Connect in 2015, targeting the same use case with vastly more resources. Balanced open-sourced their dashboard and shut down.
Agent commerce creates buyer/seller transactions that don't fit Stripe Connect's human-merchant model. FedNow enables instant settlement without card networks. Stripe Connect's pricing (0.25%+ per payout) is punitive at scale.
Agent-native payment infrastructure: programmatic escrow, instant FedNow settlement, and sub-cent fees for autonomous agent-to-agent commerce.
A micropayment platform enabling sub-dollar tips for content creators, aggregated into periodic charges. Aimed to replace advertising as the web's business model.
Card network minimums ($0.30/tx) made micropayments economically impossible. No consumer behavior pattern existed for tipping online creators in 2008.
Stablecoin rails (USDC on L2s) reduce costs to sub-cent. Creator economy normalized paying for content. AI agents need micropayment rails for API calls, data, and compute.
Stablecoin-native micropayment layer for the agent economy — sub-cent payments for both human creators and AI agents consuming services.
Open finance API aggregating data from banks, e-wallets, and lending platforms across Southeast Asia — Plaid for a region with fragmented, non-digitized banking.
SE Asian banks had no APIs, no open banking mandates, and little incentive to cooperate. Bespoke integrations across 6+ countries with different regulations burned through runway.
Indonesia, Thailand, and Philippines all launched open banking frameworks (2023-2025). LLM-powered data extraction collapses per-institution integration cost by 10x.
AI-native open finance infrastructure for SE Asia, leveraging new mandates and LLM-based connector generation to aggregate 700M+ consumers' financial data.
Banking and payment processing for legal cannabis businesses, locked out of traditional banking due to federal Schedule I classification.
Federal prohibition made banks unwilling to serve cannabis. Compliance costs were astronomical. Card networks explicitly banned cannabis transactions.
Cannabis reclassified to Schedule III (2024). SAFE Banking Act has bipartisan momentum. $35B US market, still largely cash-only. No dominant fintech player.
Purpose-built cannabis banking stack — accounts, payments, payroll, lending — for the $35B market that still can't get a bank account. First-mover window is now.
A mobile app for unbanked workers to receive wages, send remittances, and manage money without a traditional bank account.
Reaching unbanked workers required expensive physical outreach. Remittance margins were razor-thin. Competing with Western Union on trust was near-impossible.
FedNow enables instant, low-cost domestic transfers. AI multilingual interfaces solve literacy/language barriers. Employer distribution (earned wage access) provides scalable acquisition.
AI-native financial app for unbanked workers, distributed through employers. FedNow for settlement, LLM voice interface for onboarding, earned wage access as the hook.
API-first insurance infrastructure — modular underwriting, policy issuance, and claims processing embeddable into any product.
Embedded insurance was too early. Platforms didn't see insurance as a revenue line. Human underwriters were expensive. State-by-state approval outlasted runway.
Embedded insurance projected at $700B by 2030. AI underwriting replaces human actuaries. Climate risk creates urgent new products. Platforms now actively seek insurance partners.
AI-underwritten embedded insurance infrastructure with a climate-first wedge. Sell to platforms (Shopify, Uber) as a revenue add-on, not to consumers directly.
Personalized financial advice for Gen-Z — budgeting, investing basics, debt management — combining human advisors with automated tools.
Gen-Z was 15-21 in 2019 with minimal income. Human advisors cost $200+/hr, making unit economics impossible at Gen-Z price points.
Gen-Z is now 22-28, entering peak earning years. LLMs replace human advisors at near-zero marginal cost. Financial literacy demand is validated by TikTok/YouTube finance content explosion.
LLM-native financial copilot for Gen-Z's earning years. AI eliminates the human cost structure. Monetize through premium tiers and product recommendations.
Infrastructure for distributing insurance through non-traditional channels — embedded in fintech apps, e-commerce, and SaaS platforms.
Embedded distribution model was ahead of its time. Partners didn't understand insurance as a feature. Carriers were slow to approve non-traditional channels.
Embedded finance is mainstream. Climate insurance demand is surging (wildfire, flood). Carriers actively seek digital distribution as their direct channels decline.
Climate-first insurance distribution through embedded channels. Parametric climate products at point of need. The distribution unlock: platforms now want to offer insurance.
A platform for formalizing loans between friends/family — promissory notes, repayment tracking, payment facilitation for the $184B+ informal lending market.
Social stigma around formalizing personal relationships with financial contracts. Low willingness to pay. Hard to get both parties to adopt.
Housing affordability crisis made family lending for down payments mainstream necessity. AI mediation handles awkward negotiations. Credit bureau integration provides tangible value.
AI-mediated family lending focused on housing down payments. AI handles the uncomfortable parts. Frame as credit-building, not debt formalization.
Users described trading strategies in plain English; Algoriz converted them to executable algorithms and backtested against historical data.
2016 NLP couldn't reliably parse complex financial logic from natural language. The gap between described strategy and executed algorithm was too large.
LLMs genuinely understand and generate complex trading logic from natural language. Code generation can write, backtest, and debug algorithms. Retail trading participation is at all-time highs.
LLM-powered trading strategy platform — conversational strategy creation, AI-generated backtesting, autonomous execution. The NLP gap that constrained Algoriz is now addressed.
The pattern across all ten rebuild candidates is the same: AI changes the cost structure that constrained the original company. Human advisors at $200/hour become LLM agents at pennies per query. Bespoke integrations across six countries become LLM-powered connector generation. Manual underwriting becomes algorithmic.
Fintech’s next wave won’t look like the ZIRP era. It won’t be neobanks and BNPL. It will be AI agents that make financial decisions, infrastructure that settles in real time, and products that serve the markets too expensive to reach with human cost structures.
The 91 failures left behind a blueprint. At least ten are worth rebuilding — not because the ideas were wrong, but because the world finally caught up.
Methodology & Sources
Company data sourced from the yc-oss/api, an open-source mirror of Y Combinator’s public directory. Interactive charts built with D3.js + React. Analysis by Startups.RIP.
Analysis covers 812 YC companies tagged with fintech-related categories in the yc-oss/api directory as of March 2026. Keyword expansion identifies 862 total. Sub-vertical classifications based on YC tags and company descriptions. Rebuild scoring evaluates structural changes since failure, TAM trajectory, infrastructure availability, and AI impact on cost structure.
Status classifications reflect YC directory data as of March 2026. Sub-vertical classifications based on YC tags and company descriptions. Rebuild scoring evaluates structural changes since failure, TAM trajectory, infrastructure availability, and AI impact on cost structure.