In-Depth Research Report: Fintech Relies on Commissions to Win Big, DeFi Handles Trillions in Transactions but Only Makes a Small Profit

Source: Artemis & Fintech Architects

Original Title: Report: Is Fintech or DeFi a better financial system?

Translation and compilation: BitpushNews

Preface:

Fintech Architects in collaboration with the Digital Financial Think Tank Artemis have released the first KPI comparison report between Fintech and DeFi. If you’ve ever debated whether Robinhood or Uniswap is a more worthwhile asset to invest in, you’ve come to the right place.

This report is the first to compare fintech stocks and crypto tokens on the same level. Covering payments, digital banking, trading, lending, and prediction markets, the report compares revenue, user numbers, take rates, industry KPIs, and valuation metrics. The results are astonishing:

• Hyperliquid’s trading volume has exceeded 50% of Robinhood’s;
• DeFi protocol Aave’s total outstanding loans surpass the buy-now-pay-later giant Klarna;
• Growth speed of stablecoin settlement networks far outpaces traditional payment providers;
• Wallets like Phantom and MetaMask have user bases large enough to rival digital banking giants like Nubank and Revolut.

We find that valuation fully reflects this game: crypto assets are either highly overvalued or deeply discounted based on their expected monetization potential. Ultimately, we believe the core issue of integration is: will the crypto industry learn to build “toll booths,” or will the fintech industry adopt the “open rails” of crypto?

The Battle of Two Financial Systems

For years, we have viewed cryptocurrency and fintech as parallel universes. One is a regulated, audited system traded on Nasdaq; the other is permissionless, traded on decentralized and centralized exchanges. They share a common language: revenue, trading volume, payments, lending, trading, but with different “accents.”

This situation is changing. With Stripe acquiring Bridge, Robinhood launching prediction markets, and PayPal issuing its own stablecoin, the boundaries are blurring. The question is, when these two worlds collide, how does their strength compare?

Chart explanation: In our chart, purple represents cryptocurrencies (Crypto), green represents equity companies (Equities). Currently, Robinhood ranks first in trading volume, but the second place is Hyperliquid……

We decided to conduct an experiment: select well-known fintech companies in payments, digital banking, BNPL, and retail brokerage, and stack them against crypto-native protocols. Using the same metrics (P/S, ARPU, TPV, user count), green bars represent US stocks, purple bars represent token protocols.

A panoramic view of these two financial systems emerges: on-chain financial protocols often match or surpass fintech competitors in trading volume and asset scale, but their captured economic benefits are only a small fraction of the latter. In contrast, crypto asset valuations are either extremely high or very low, with little middle ground. Moreover, their growth rates are on completely different scales.

Payments: The Pipelines of Capital Flow

Starting with the largest category in fintech—funds transfer.

Green camp (major players):

• PayPal’s annual transaction volume reaches $1.76 trillion.
• Adyen processes $1.5 trillion.
• Fiserv (that almost-forgotten infrastructure layer) processed $320 billion.
• Block (formerly Square) drove $255 billion via Cash App and merchant network.

Purple camp (Artemis-estimated annual B2B payment volume):

• Tron moved $68 billion in stablecoins.
• Ethereum reached $41.2 billion.
• BNB processed $18.6 billion.
• Solana approximately $6.5 billion.

In absolute numbers, they are not on the same scale. The stablecoin transfer volume of all mainstream public chains is only about 2% of traditional fintech payment processors. In market share charts, purple bars are almost negligible.

But interestingly, the growth rates are notable:

Last year, PayPal’s total payments grew by only 6%, Block by 8%, while Europe’s darling Adyen achieved 43% growth (which is very strong by fintech standards).

Looking at blockchain: Tron grew by 493%, Ethereum by 652%, BNB by 648%, and Solana fastest at 755% YoY growth. It’s important to note this data is estimated by Artemis based on McKinsey research for B2B payments.

Conclusion: Growth in stablecoin rails far exceeds traditional payments, despite starting from a much smaller base.

Now, who makes more money? Fiserv takes a 3.16% cut per transaction, Block takes 2.62%, PayPal 1.68%. Even low-margin operators like Adyen take a 15 basis point cut.

As for blockchain, their take rates on stablecoin and asset transfers are extremely low, around 1 to 9 basis points. Blockchain operates by collecting Gas fees, bypassing interchange and merchant fees, achieving huge efficiency advantages over traditional channels. While this limits protocol revenue, it creates profit margins for upper-layer payment orchestrators.

Digital Banking: Wallets as New Bank Accounts

On the fintech side, we have licensed banks: Revolut, Nubank, SoFi, Chime, Wise.

On the crypto side, we see wallets and yield protocols: MetaMask, Phantom, Ethena, EtherFi. Though they don’t carry the “bank” name, millions of people store assets there and earn interest.

User comparison:

• Nubank has 93.5 million monthly active users (MAU), the largest digital bank globally.
• Revolut has 70 million users.
• MetaMask follows with 30 million MAU, surpassing Wise, SoFi, and Chime.
• Phantom has 16 million MAU and has expanded to multi-chain, even launching its own debit card and tokenized stocks.

Deposit scale (funds deposited):

• Revolut holds $40.8 billion in customer balances.
• Nubank has $38.8 billion.
• SoFi has $32.9 billion.

In crypto, EtherFi (liquidity re-staking) holds $9.9 billion, Ethena (synthetic USD) $7.9 billion. Although called TVL (Total Value Locked) industry-wide, from a user perspective, it’s the money stored somewhere to earn yields.

Profitability gap:

• SoFi’s ARPU (average revenue per user) is $264/year, thanks to cross-selling loans, investments, credit cards.
• EtherFi’s ARPU is also high at $256, comparable to SoFi. But the awkward part: EtherFi has only 20,000 active users, while SoFi has 12.6 million. This means DeFi protocols can efficiently extract value from niche users like top digital banks but have yet to reach the masses.
• By contrast, MetaMask generated about $85 million last year, with an ARPU of only $3.

Valuation logic:

Market valuations for both are surprisingly similar. Revolut’s P/S ratio is 18x, EtherFi’s is 13x, Ethena’s is 6.3x. The current trend is “wallet banking”: MetaMask adding debit cards, Phantom integrating fiat channels.

Trading: On-Chain DEX Challenging Traditional Brokers

In capital markets, on-chain exchanges are astonishingly large.

• Robinhood processed $4.6 trillion in trading volume over the past 12 months.
• Hyperliquid (decentralized perpetual contracts) handled about $2.6 trillion.
• Coinbase processed $1.4 trillion.
• Mainstream DEXs like Uniswap and Raydium now match Coinbase’s trading volume. This was unimaginable three years ago.

However, the “DEX paradox” lies in take rates:

• Robinhood’s overall take rate is 1.06%, Coinbase’s is 1.03%.
• DEXs’ take rates are between 3 and 9 basis points.

Take rate = LTM revenue / trading volume. Revenue for eToro, Coinbase, Robinhood, Bullish comes from financial reports. Raydium, Aerodrome, Uniswap, Meteora data from Artemis.

This means: Uniswap’s revenue from $1 trillion in volume is only about $29 million; Coinbase’s from $1.4 trillion is $1.4 billion.

In valuation terms, the results align with these figures:

• Coinbase trades at 7.1x sales
• Robinhood’s trading price is 21.3x, high for brokers but supported by growth
• Schwab at 8.0x, mature business multiple
• Uniswap at 5.0x fees
• Aerodrome at 4.8x fees
• Raydium at 1.3x fees

Markets do not value these protocols as high-growth tech companies, partly because their take rates are lower compared to traditional brokers.

Market cap / LTM revenue. Publicly reported stock market caps are from Yahoo Finance, tokens from Artemis.

The stock price performance chart shows market sentiment.

Since late 2023, Robinhood has risen about 5.7x, riding the wave of retail investing and crypto revival. Coinbase increased by 20% in the same period. Uniswap fell 40%. Despite large trading volumes flowing into their DEXs, tokens have not captured as much value, partly because their use as investment tools is unclear. The only exception is Hyperliquid, which, due to its rapid rise, nearly matched Robinhood’s gains in the same period.

Although DEXs have historically failed to capture value and are seen as public goods, projects like Uniswap are turning on their “fee switches,” using fees to burn UNI tokens, now generating $32 million annualized revenue.

We hope that as more trading shifts on-chain, value can flow back into DEX tokens. Hyperliquid is a good example of success. But currently, before mechanisms like Hyperliquid’s value capture are in place, DEX tokens will underperform their CEX stock counterparts.

Lending: Underwriting the Next Generation

Lending becomes an even more interesting comparison. On one side, you have core fintech lending products: unsecured consumer credit.

• Affirm allows you to pay for Peloton bikes in four installments
• Klarna does the same for fast fashion
• Lending Club pioneered P2P lending before transforming into a true bank
• Funding Circle underwrites small business loans

These companies profit by charging borrowers higher fees than they pay depositors, hoping defaults won’t eat into the interest margin.

On the other side, you have collateralized DeFi lending: Aave, Morpho, Euler. Borrowers deposit ETH, lend USDC, and pay algorithmically determined interest rates. If collateral drops too much, the protocol automatically liquidates. No collection calls or write-offs. These are fundamentally different from traditional loans.

Starting with the loan book: Aave has $22.6 billion in outstanding loans. That’s more than Klarna ($10.1B), Affirm ($7.2B), Funding Circle ($2.8B), and Lending Club ($2.6B) combined. The largest DeFi lending protocol has a bigger loan book than the biggest BNPL player. Let that sink in.

Total loans of Lending Club, Funding Circle, Affirm, Klarna, and Figure are from financial reports. Loans and deposits for Euler, Morpho, Aave are from Artemis.

Morpho added another $3.7 billion. Euler, after a 2023 hack, was restarted and holds $861 million. The DeFi lending stack, in total scale, can rival the entire listed digital lending sector within about four years. But the economic model is inverted.

Funding Circle’s “net interest margin” (NIM) is 9.35% (due to its private credit-like business model). Lending Club’s is 6.18%. Affirm, despite being a BNPL firm rather than a traditional lender, earns 5.25%. These are fat spreads, compensating for credit risk absorbed through actual underwriting.

In crypto, Aave’s net interest margin is only 0.98%. Morpho’s is 1.51%. Euler’s is 1.30%. Despite larger loan books, DeFi protocols’ yields are generally lower than those of fintech lenders.

Aave, Euler, Morpho’s net interest margin = revenue / loans. Stock net interest margin is from financial reports.

DeFi lending is designed to be over-collateralized. To borrow $100 on Aave, you typically need to deposit $150 or more in collateral. The protocol bears no credit risk. It bears liquidation risk—borrowers pay for leverage and liquidity, not for unborrowable credit privileges.

Fintech lenders do the opposite. They offer unsecured credit to consumers wanting buy-now-pay-later. The interest spread compensates for those who never repay. This is reflected in actual default loss figures, and managing these losses is core to underwriting.

Stock credit loss ratios come from public financial reports.

So, which model is better? It depends on what you optimize for.
Fintech lending serves borrowers who need money now and takes on real underwriting risk. It’s also brutal. Early digital lenders (OnDeck, Lending Club, Prosper) nearly went bankrupt multiple times. Despite operational success, Affirm’s stock has fallen about 60% from its peak, often because underwriting revenue is SaaS-like and doesn’t fully account for inevitable future losses.

DeFi lending is a leverage business. It serves those who already have assets but want liquidity without selling—similar to margin accounts. Aside from collateral quality, there are no credit decisions. It’s capital-efficient, scalable, and earns tiny profits on huge volumes. It’s only useful for those with substantial on-chain assets seeking yield or leverage.

Prediction Markets: Who Knows?

Finally, let’s look at prediction markets.

These projects are the newest battleground between fintech and DeFi, and also the most peculiar.

For decades, they were mostly academic curiosities, favored by economists but disliked by regulators. Iowa Electronic Markets once ran small election predictions. Intrade briefly thrived before shutting down. Most such projects were labeled “gambling” or “sports betting.”

The idea that you can trade on real-world outcomes, and that these markets can produce more accurate forecasts than polls or experts, was largely theoretical.

All this changed in 2024, accelerating during the second Trump administration. Polymarket handled over $1 billion in election bets. Kalshi won a lawsuit against the US Commodity Futures Trading Commission (CFTC) and launched political contracts for US users. Robinhood, never missing a trend, added event contracts. Meanwhile, DraftKings, a giant already running a de facto prediction market through daily fantasy sports, quietly made $5.5 billion in revenue with a $15.7 billion market cap.

Artemis Prediction Market Dashboard

(Chart note: Spot trading volume data for Kalshi and Polymarket from Artemis. For DraftKings, trading volume is based on “Sportsbook Handle,” i.e., total settled customer bets in its sports betting products.)

This category moved from niche to mainstream in about 18 months, with weekly prediction market trading volume reaching around $7 billion, hitting a new all-time high.

In the past 12 months, DraftKings processed $51.7 billion in trading volume. Polymarket handled $24.6 billion, about half of the former, despite being a crypto-native protocol that technically doesn’t allow US users. The regulated US alternative, Kalshi, processed $9.1 billion. In terms of trading volume alone, Polymarket is highly competitive. While Kalshi is still fighting in court, it has already built a liquid global prediction market on Polygon.

But when it comes to revenue, the comparison doesn’t hold.

DraftKings generated $5.46 billion last year, while Kalshi only brought in $264 million. Polymarket, after introducing taker fees on its “15-minute crypto market,” has an annualized revenue run rate of only $38 million.

Prediction market revenue comparison (Chart note: Polymarket revenue data from Artemis, Kalshi revenue from cited sources, DraftKings 12-month (LTM) revenue from financial disclosures.)

The huge gap lies in take rates, or “hold” in sports betting terminology:

• DraftKings retains 10.57% of each dollar wagered. This is typical sports betting: the bookmaker takes a cut, offers odds, and manages risk.
• Kalshi takes 2.91%, a thinner profit margin more suited to financial exchanges.
• Crypto-native Polymarket only takes 0.15%. With $24.6 billion in volume, it captures very little value.

Take rate = last 12 months’ revenue / trading volume

This echoes the dynamics of decentralized exchanges (DEXs). Polymarket’s focus isn’t on value capture but on providing infrastructure for prediction markets—matching buyers and sellers and settling contracts on-chain. It doesn’t set odds, manage assets, or act as your counterparty. While highly efficient, monetization isn’t its core focus.

However, investors clearly believe Polymarket can eventually monetize:

• Polymarket’s valuation is $9 billion, with a P/S ratio of 240x.
• Kalshi’s valuation is $11 billion, with $264 million in revenue, a 42x multiple.
• DraftKings trades at only 2.9x revenue.

Venture capital (VC) keeps pouring money into these platforms, while “traditional” operators like DraftKings and Flutter (FanDuel) watch their stock prices plummet.

Prediction Market Valuation Comparison

(Chart note: Valuations for Kalshi and Polymarket are based on latest private funding rounds. DraftKings market cap from Yahoo Finance.)

Polymarket’s valuation assumes it will either monetize in a grand way or evolve into something much larger than a prediction market. At over 200x P/S, you’re not buying a company but a call option on a new financial primitive. Maybe Polymarket becomes the default venue for hedging any real-world event. Maybe it adds more sports, earnings reports, weather, or anything with binary outcomes. Maybe it captures a larger percentage than 0.15%, suddenly generating billions in revenue.

This is the purest form of the “convergence” problem: will the future belong to regulated exchanges with take rates and compliance departments, or to permissionless protocols that let anyone, anywhere, bet on anything without leaving profits for the house?

The Final Fusion

A few years ago, we couldn’t compare DeFi and Fintech directly. But now, the data is in front of us.

Cryptocurrencies have built a financial infrastructure capable of rivaling fintech in trading volume, user base, and asset scale. Stablecoin rails are more globalized than traditional payments, Aave’s ledger is larger than Klarna’s, and Polymarket is eating into the betting market share. The technology is in place, and products have found audiences.

But there’s a key “trap”: the crypto industry’s ability to capture economic value (take rates) is far inferior to traditional fintech.

You can see this as a “feature”: it represents the ultimate democratization and efficiency of financial services, destroying profit margins to benefit users. Or as a “bug”: if protocols can’t generate enough revenue, their token value sustainability will be challenged.

Fusion is happening. Banks are piloting tokenized deposits, NYSE is researching tokenized stocks, and the total market cap of stablecoins has surpassed $300 billion. Fintech giants see the future—they won’t sit idly by; they will absorb and assimilate it.

The question for the next decade is simple: will the crypto world learn to build toll booths, or will traditional finance leverage the roads of crypto? We bet both will happen.