Skip to main content
Back to blog
  • Fintech

Updated 16 min read

Fintech trends 2027. 10 shifts from 208 fintechs and banks

Ranked by how many reports from McKinsey, BCG, Deloitte, Accenture, PwC, Visa and Mastercard carry each one, with what to do in 2027

Emily Carter

Fintech and banking

ShareLinkedInXEmail

In 2018, a quarter of new US checking customers simply picked their existing bank, without shopping around. McKinsey now puts that share at 4%.

That is one of ten shifts that kept coming back when we read 65 banking and payments reports. McKinsey, BCG, Deloitte, Accenture, PwC, Visa and Mastercard published them between October 2025 and October 2026. The biggest bank survey among them, Accenture’s, asked payments and technology leaders at 208 financial institutions in 17 markets.

Between them, the ten decide where your deposits sit, who holds your customer’s main account and how much of your cost base you can still defend in 2027.

We ranked them by how many reports carry each one, with each shift’s strongest numbers and what to do about it next year. We also kept the places where the firms disagree. On one 2030 forecast, two of them are more than five times apart.

Number one is the only shift that all seven firms carry.

Sign up with email

A quick word on method. A shift counts for a report when it is one of that report’s 2 to 4 main themes. The counts describe this library, not the whole market: McKinsey wrote 17 of the reports and Visa 14, while Accenture wrote 3. We checked every number below on its page in the source report.

The short version

  1. AI agents start doing real bank work. In corporate and investment banking, BCG expects AI to free 25% to 40% of banker capacity within five years.
  2. Stablecoins grow fast, but real payments stay small. About $390 billion in 2025, McKinsey estimates.
  3. Money moves on-chain beyond stablecoins. Big banks already move more than $4 trillion a year through tokenized deposits (McKinsey).
  4. AI agents will shop and pay. Accenture sees about $3.1 trillion of online commerce running through agents by 2030.
  5. The cost gap is wide. Many traditional banks run a cost-to-income ratio above 60%, against about 35% at well-run digital banks (BCG).
  6. Everyone uses AI, few get value. Only 4 of 50 banks reported realized ROI from AI in 2025 (Deloitte, citing Evident).
  7. Trust decides adoption. Only one in ten consumers would let an agent finish a purchase alone (Mastercard).
  8. AI makes fraud bigger. 78% of payments leaders expect fraud to rise significantly with agentic payments (Accenture).
  9. The main account is loosening. Only 4% of new US checking customers pick their existing bank without shopping around, down from 25% in 2018 (McKinsey).
  10. Business clients move first. B2B makes up about 60% of real stablecoin payments (McKinsey).
Figure 1. The 10 shifts by number of reports that carry them (of 65)
ShiftReports
1. AI agents do real bank work19
2. Stablecoins grow, real payments stay small18
3. Money moves on-chain beyond stablecoins13
4. AI agents shop and pay12
5. The cost gap12
6. Everyone uses AI, few get value12
7. Trust and consent decide adoption12
8. AI makes fraud bigger and faster10
9. The main-bank relationship loosens10
10. Business clients move first10

Source: Humbleteam reading of 65 reports from McKinsey, BCG, Deloitte, Accenture, PwC, Visa and Mastercard, October 2025 to October 2026.

1. AI agents are starting to do real bank work

19 reports, all 7 firms

The language changed this year. Fewer reports talk about chatbots and copilots. More describe agents that run a whole process, such as KYC checks, servicing, collections, coding and back-office operations, while people supervise.

The capacity numbers are big. In corporate and investment banking, BCG’s Future of Finance team expects AI to free roughly 25% to 40% of banker capacity and 20% to 35% of operations capacity over the next five years. McKinsey describes set-ups where one person manages roughly 20 to 30 agents.

Accenture ran the numbers for the top 200 banks. If they scale gen AI over the next three years, revenue could rise 5%, operating costs fall 8% and loan-loss provisions fall 16%. That adds up to $289 billion.

BCG says the value will come from redesigning core workflows, such as credit, onboarding and servicing. A tool added on top of the old workflow will not deliver it.

What to do in 2027: pick one process you can rebuild from end to end, such as onboarding or collections. Give one person ownership of the outcome and of the agents that deliver it. Measure it in cost per case and time to decision.

2. Stablecoins grow fast, but real payments stay small

18 reports, 6 firms (all but Accenture)

The rules have arrived. The US GENIUS Act comes up in 13 of the reports and the EU’s MiCA in 9. Supply is growing fast, but most activity is still trading. Mastercard’s Payments Navigator, citing BCG, says up to 95% of stablecoin activity is still tied to crypto trading and decentralized finance.

Real payments are small, but they doubled. McKinsey’s payments team puts real stablecoin payments at about $390 billion in 2025, roughly 0.02% of global payments. That is more than double the 2024 level.

Banks are moving anyway. In Visa’s survey of commercial banks, 58% are either piloting stablecoins (21%) or planning to be first movers (37%). The reason is deposits. Deloitte estimates that payment stablecoins could put more than $1 trillion of bank deposits at risk.

What to do in 2027: decide your role. Issue your own, join a group of banks, or use someone else’s rails. Model what a deposit outflow would do to your funding. Then start where stablecoins already work: B2B and cross-border payments (see shift 10).

3. Money and assets are moving on-chain, beyond stablecoins

13 reports, 6 firms (all but Mastercard)

Stablecoins get the headlines. Underneath, the reports point to a wider layer of on-chain money: tokenized deposits, tokenized funds and real-world assets, and central bank digital money.

The bigger flow is already inside banks. McKinsey estimates that the largest global banks move more than $4 trillion a year through tokenized deposit systems. That is an order of magnitude more than stablecoin payments.

Waiting is expensive. BCG’s digital assets team puts up to 15% of bank revenues and 30% of profits at risk by 2035.

Public money is coming, slowly. PwC Italy notes that the digital euro is planned for 2029. In a Strategy& survey it cites, 37% of respondents had never heard of it.

What to do in 2027: treat tokenized deposits as the bank’s own answer to stablecoins. Pick one institutional use case, such as treasury or intraday liquidity, and run it on-chain.

4. AI agents will shop and pay for customers

12 reports, 6 firms (all but PwC)

Agents will search, compare, choose and pay. That raises questions banks, networks and acquirers have not answered yet. Who is the agent? What did the customer agree to? Who pays when it goes wrong?

The forecasts are large. Accenture estimates that more than 30% of online commerce could run through AI agents by 2030, close to $3.1 trillion in transactions. McKinsey’s baseline puts about $75 billion of global payments revenue at risk from agentic AI by 2030.

Readiness is low. In Visa Acceptance’s survey of acquirers in Brazil, the UAE and the US, only 23% plan to speed up merchant readiness. The other 77% are waiting for demand or for large merchants to move first.

What to do in 2027: make your products and terms readable by agents, as well as by people. Decide how an agent proves who it is and what it may spend. Rewrite your dispute rules for a purchase no human clicked.

5. The cost gap: 30-40% cuts on the table

12 reports, 6 firms (all but Mastercard)

Incumbents still run far heavier than digital banks. Most reports pitch AI first as a way to close that gap.

The gap in two numbers. BCG finds many traditional banks stuck at a cost-to-income ratio above 60%, against about 35% at well-run digital banks. McKinsey’s UK team counts about 200 to 250 employees per million customers at leading neobanks. Large incumbents need roughly 700 to 1,100.

The promised cut. BCG, writing with OpenAI, says AI agents could cut bank costs by 30% to 40% by 2030.

The catch. AI has its own bill. Deloitte found that 56% of financial services finance leaders expect AI costs and complexity to rise substantially through 2027.

What to do in 2027: track cost per customer, not the number of AI projects. Put the cost of running AI, including models, data and oversight, in the same budget line as the savings it claims.

6. Everyone uses AI, few get value at scale

12 reports, 5 firms (McKinsey, BCG, Deloitte, Accenture, Visa)

Adoption is close to universal. Measured returns are rare.

The numbers agree. Deloitte, citing Evident, notes that only 4 of 50 banks reported realized ROI from AI use cases in 2025. McKinsey finds that 88% of companies use AI in at least one function, but just 7% have deployed it fully. BCG’s wider study found that 5% of companies get value at scale from AI, while 60% do not.

In Asia, nearly 80% of the financial institutions McKinsey works with use AI. Yet a similar share worldwide report no significant impact on profit.

The fix is mostly people. Klemens Hjartar, a senior partner at McKinsey, puts the change at 85% people and culture and 15% technology.

What to do in 2027: run fewer use cases and rebuild each one fully: process, data, roles and incentives. Report AI results in the P&L, not as a count of pilots.

12 reports, 5 firms (McKinsey, Deloitte, Accenture, Visa, Mastercard)

People like help from AI. They do not yet hand it their money decisions.

Consumers hold back. Mastercard finds that only one in ten consumers would let an agent complete a purchase on its own. Accenture puts the share willing to let agents handle payment choices at 12%, and 87% of payments leaders call trust the main barrier.

Deloitte reports that sentiment even went backwards. In 2025, 70% of consumers had some comfort with agents making purchases. By June 2026, 55% were uncomfortable.

Banks start with an edge. In McKinsey’s survey, 62% of consumers most trust their primary bank to offer gen AI financial services, against 19% for a major tech company. Visa finds that US intent to use stablecoins rises from 36% to 56% once bank-level fraud protection and deposit insurance are added.

What to do in 2027: design consent as a product. Give customers clear spending limits, controls they can see and an easy way to undo. This is the shift where the interface decides whether people trust the agent.

8. AI makes fraud bigger and faster

10 reports, 5 firms (Deloitte, Accenture, Visa, Mastercard, PwC)

AI-made scams, deepfakes and fake agents scale faster than defenses built in silos.

Leaders expect it to get worse. Accenture found that 78% of payments leaders expect fraud to rise significantly with agentic payments. (Mastercard quotes the same figure, so count it once.) Deloitte projects that US losses from authorized push payment fraud could reach nearly $15 billion by 2028.

Customers already feel it. In Visa’s remittance survey, one in four remitters worldwide report fraud exposure. That rises to 40% in India and 36% in the US.

What to do in 2027: bring fraud and cyber into one real-time view, as Visa’s consultants recommend. Agree who is liable when an agent pays the wrong party, before the first case reaches a customer.

9. The main-bank relationship is loosening

10 reports, 4 firms (McKinsey, Visa, Mastercard, PwC)

Customers spread their money across more providers. Deposits chase yield. AI agents may soon move idle cash for the customer.

Loyalty is thinning. McKinsey finds that only 4% of new US checking customers choose their existing bank without shopping around, down from 25% in 2018. In six European markets, customers now keep accounts at 2.6 banks on average, up from two in 2021. In the UK, neobanks and fintechs already hold about 27% of banking relationships.

Agents could speed it up. McKinsey estimates that if just 5 to 10 percent of UK balances moved to higher-yield accounts through AI agents, deposit profits could fall by 20% or more.

Business clients drift too. Three in four commercial banks in Visa’s survey say at least 26% of their commercial clients are moving funds away from their main bank.

What to do in 2027: be among the first few banks a customer considers. McKinsey finds that is where the choice now gets made. Then use your data to make the main account worth opening every week of the month.

10. Business clients move first

10 reports, 4 firms (McKinsey, BCG, Visa, Mastercard)

Many of these shifts show up with businesses before consumers.

B2B leads on-chain. McKinsey puts B2B at about $226 billion a year, roughly 60% of real stablecoin payments.

Corporate loyalty is earned every year. McKinsey finds that corporate, investor and sponsor clients reallocate 12 to 25 percent of their banking wallet each year.

The offer lags. Only 2% of commercial banks in Visa’s survey think their cross-border offering is good enough. 95% say the speed of payment rails hurts their clients. Mastercard, citing Juniper Research, notes that only 2-3% of B2B payments by value ran on card rails in 2025.

What to do in 2027: launch new payment products with business clients first. Fix cross-border speed and treasury integration, because that is where clients are already moving.

Where the reports disagree

How big agentic commerce gets. Accenture and Deloitte both land near 30% for 2030, then part ways. Accenture means more than 30% of online commerce running through agents, about $3.1 trillion. Deloitte means roughly 30% of digital commerce value influenced by agents, and $17.5 trillion of merchandise value “enabled” by them. It is a definition gap: transactions an agent runs versus purchases it touches. In a board pack, I would use the narrower number.

Figure 2. Agentic commerce by 2030: two forecasts ($ trillion)
ForecastValue
Accenture: online commerce run through agents3.1
Deloitte: merchandise value enabled by agents17.5

Source: Accenture, “Agentic commerce rewrites payment choice”, May 2026; Deloitte, “Future of payments: Insights for 2027”, 2026.

How fast payments revenue grows. McKinsey and BCG both see growth slowing. McKinsey sees it moderating from 9% to about 4% a year. BCG expects 5% a year to 2030, down from 7%. They agree on the direction, not on how far growth falls.

Whether AI saves money or costs it. BCG’s 30% to 40% cost cut sits next to Deloitte’s finding that 56% of finance leaders expect AI costs to rise substantially. They don’t cancel out. The difference is whether a bank rebuilds the process or adds tools on top.

Whether customers are warming to agents. Deloitte reports comfort with agent purchases going backwards between 2025 and June 2026. The forecasts in shift 4 assume that turns around.

Just below the top 10

Five themes came close:

  • Legacy core and data hold everything back (8 reports, 6 firms). McKinsey finds that at many large banks, 60 to 80 percent of technology spending goes to running existing systems.
  • Banks at peak profit while the base erodes (8 reports, 3 firms). McKinsey: bank net income reached $1.3 trillion in 2025, up 7% on the 2024 record.
  • Macro and geopolitics (8 reports, 5 firms), mostly the economic outlooks.
  • Payments growth slows (7 reports, 3 firms). See the McKinsey and BCG numbers above.
  • Fintech comes of age (5 reports, 4 firms). McKinsey counts about $650 billion of fintech revenue in 2025, up about 21%. BCG measures growth of 22%, more than four times as fast as incumbents.

What I’m watching in 2027

Three things would change this list for me. The first is a real test of agents paying at scale, disputes included. The second is a deposit outflow to stablecoins that shows up in a bank’s results. The third is the AI cost line in bank budgets, set next to the savings it was meant to bring.

Banks sometimes ask us what they should run in 2027. After these 65 reports, our answer got shorter. Pick the two or three shifts your customers will feel first, and rebuild those properly.

Thank you to the research teams at all seven firms. Their work made this list possible.

Sources

Read next

Let's talk

Write Sergey on LinkedIn
Write Alexandra on LinkedIn
Write Emin on LinkedIn
Have questions? Ask AI
ChatGPT
Claude
Perplexity
Opens a new chat with context about us pre-loaded — ask anything

We’ll reply within 24 hours with case studies, a timeline, and an estimate.

Prefer email? Write to hi@humbleteam.com
All set – our team’s on it. Expect a reply soon.
Send another one
Oops! Something went wrong while submitting the form.
Have questions? Ask AI
ChatGPT
Claude
Perplexity
Opens a new chat with context about us pre-loaded — ask anything
Europe
Národní 135/14, Prague
Middle East
UAE, Dubai, Internet City Offices
Services by industryCompare usHealthtechSaaSDubaiLondonBlogFor AI SaaS
We use cookies to enhance your browsing experience,
serve personalised ads or content, and analyse our traffic.
By clicking "Accept All", you consent to our use of cookies.
Accept all
Decline
Privacy policy