In 2024, global banks posted $1,200 billion in net profits. An absolute record across all industrial sectors combined. And yet, markets value the banking sector at a price-to-book ratio of 1.0, which is 70% below the cross-sector average. This paradox says everything. Markets do not believe in the sustainability of this model. They see what many players still refuse to admit: past performance was boosted by high rates and a favorable economic climate. The real value drivers for the coming decade are already at work elsewhere. They are called: agentic AI, tokenization of real-world assets, embedded finance, stablecoins, decentralized infrastructures. And they are not just optimizing the existing system: they are rebuilding its foundations.
The system is at its peak, but under structural pressure
Three headwinds are blowing at the same time. The normalization of rates is eroding the interest margins that fueled record profits. Customer loyalty is collapsing: in 2024, only 4% of new credit card holders in the United States chose their existing bank without exploring alternatives, compared to 10% in 2018. And competitive pressure from neobanks / fintechs, private credit managers, and investment platforms is systematically capturing the most profitable segments.
Meanwhile, banks collectively spend at least $600 billion per year on technology without generating proportional productivity. The question is no longer budgetary. It is strategic. Regarding infrastructure, the number one cause of friction cited by European financial institutions remains the dependence on technological architectures from the 1970s-1990s; COBOL still running in production, batch systems that cannot compete with neobank platforms, and compounded technical debt where every new regulation becomes a workaround rather than a capability. Revolut has 40 million customers. Nubank has 90 million. These figures are no longer warning signals: they are war bulletins.
The first revolution: AI no longer just answers, it acts
Agentic AI is the most immediate and radical disruption underway in 2026. It no longer settles for answering questions: it orchestrates complex workflows, takes real-time decisions, and acts without human intervention. The numbers are staggering. 92% of leading fintechs integrated at least one autonomous agent into production in Q1 2026. 85% of retail banks deployed agents capable of resolving disputes and processing credit applications without a human, compared to only 40% in early 2025. 44% of finance teams will use agentic AI this year, a +600% increase in one year.
Concrete applications already exist on a large scale. JPMorgan’s COIN replaces 360,000 hours per year of legal contract review. McKinsey projects a reduction in operational costs of 15 to 20% net for adopting banks, representing $700 to $800 billion in potential savings industry-wide. KPMG estimates the impact on corporate productivity at $3,000 billion annually, with an average EBITDA improvement of 5.4%.
But agentic AI is on both sides of the ring. If only 5 to 10% of the $23,000 billion in current account deposits migrate toward better rates, orchestrated by consumer-side AI agents, it could reduce deposit profits by 20% or more. McKinsey evaluates the potential loss for immobile banks at $170 billion. AI turning against its passive users is not a metaphor. It is a financial projection. It is also worth noting the risks related to AI. Ill-prepared financial institutions risk security flaws and confidentiality breaches, further challenging their expertise in cybersecurity and compliance. Using agentic solutions is not inherently reassuring and can be costly in the end, even though secure solutions exist.
The second revolution: tokenization turns every asset into programmable infrastructure
Tokenization consists of representing real-world assets—bonds, real estate, commodities, money market funds—on a blockchain, making them fractionalizable, tradable 24/7, and usable as collateral in real-time. It has moved from experimentation to operational infrastructure. The tokenized RWA market (excluding stablecoins) grew by 266% over 2025 to reach $24 billion in February 2026. This figure seems modest, but compared to $500,000 billion in TradFi assets, it represents the start of a historic transfer. Institutional consensus projects between $16,000 and $30,000 billion of tokenized assets by 2030, or 5 to 10% of global GDP, with a CAGR of 53%.
Institutional moves are already decisive: BlackRock integrated its BUIDL fund ($2.4 billion AUM) directly into DeFi protocols via UniswapX, the first native DeFi integration for the world’s largest asset manager. It simultaneously acquired UNI tokens, becoming a decentralized governance player. Its internal plan reportedly aims to tokenize its entire iShares franchise—1,700 funds—within the next 3 to 12 months. JPMorgan processes over $1 billion per day via its JPMD deposit token on Ethereum L2. Citi Token Services processes $200 million daily in cash management. Goldman Sachs and BNP Paribas are testing tokenized bonds and gold via the Canton Network with 30 partner institutions.
This is no longer a research thesis. It is production infrastructure. Additionally, next-generation « Web3 » infrastructures are arriving, and many current problems will find solutions, accelerating adoption.
The third revolution: stablecoins become the plumbing of global finance
Stablecoins crossed a historic threshold: $320 billion in capitalization in March 2026. In January of this year, stablecoin networks moved over $10,000 billion in transaction volume, a figure that now rivals traditional settlement systems like Visa. In 2024, stablecoins already represented $2,000 billion in transactions over 138 million transfers. They serve as a B2B settlement layer, for treasury flows, and cross-border payments, bypassing the delays and opaque fees of SWIFT. ECB analysts project their capitalization could exceed $1,000 billion by the end of 2026 or 2027.
In response, the ECB is preparing. The digital euro (CBDC) entered the technical preparation phase in October 2025, with an agreement from European Finance ministers on a holding limit framework. If legislation is adopted in 2026, a pilot is conceivable in 2027 and initial issuance in 2029. The explicit goal: preserve European monetary sovereignty against private US stablecoins and foreign CBDCs. The real performance comparison is brutal. Traditional SWIFT settles in T+1 to T+3 with high and opaque fees, without programmability. Stablecoins settle in seconds, at near-zero cost, with programmable smart contracts. Already operational at scale.
The fourth revolution: embedded finance erases the border between bank and platform
73% of consumers now prefer to buy financial products within non-banking applications. This is the deepest signal of the shift in the balance of power. Embedded Finance represents the integration of financial services (loans, insurance, payments…) directly into non-financial platforms. Amazon, Apple, Shopify, and Uber integrate financial capabilities by capturing the customer relationship without being regulated as banks.
The global embedded finance market is estimated between $86 and $156 billion in 2026, on a trajectory toward $7,000 billion by 2030, a CAGR of 23 to 32%. BaaS (Banking-as-a-Service) reduces the time-to-market for new financial products by 60%. And an embedded customer generates up to 5 times more revenue for the platform than a traditional customer. Banks have a choice: be the invisible back end (BaaS, APIs) or be the platform (super app). Both positions are viable. What is no longer viable: staying invisible in a landscape where 73% of financial transactions flow outside of banking apps.
Tradfi and defi: convergence is underway, not war
Traditional finance and decentralized finance were long opposed as if one had to kill the other. The reality of 2026 is more nuanced and interesting. At the 2026 Davos Forum, the debate was no longer about « if » but « how » to merge $500,000 billion of TradFi assets with blockchain rails. Both systems bring irreplaceable structural advantages. TradFi brings deep capital, institutional liquidity, regulatory legitimacy, and established trust infrastructures. DeFi brings contract programmability, 24/7 availability, instant settlement, superior capital efficiency, global accessibility… and many other innovations to come.
The result in 2026 is a hybrid system: institutions tokenizing their assets to deploy them on regulated DeFi rails, and DeFi protocols institutionalizing by integrating permissioned KYC/AML layers. In Europe, MiCA—certainly rigid and not facilitating innovation—entered into force on January 1, 2025, providing a clear framework: 53 CASP licenses have already been granted by European regulators, offering a single passport to operate across the 27 member states. DeFi is no longer a speculative wild west. Among the top 50 DeFi protocols, each now exceeds $1 billion in TVL. Global TVL is heading toward $300 billion in 2026.
The obstacles are real and underestimated
No honest analysis can ignore the persistent risks. Technically: $1.6 billion lost to DeFi security breaches in 2025. 90% of exploits involve oracle manipulation—the bridges between the real world and smart contracts. Composability is the sector’s great strength and its main systemic risk: a failure in one protocol can destabilize the entire interconnected ecosystem. RWA side: the value of a building or a debt cannot be determined on-chain. It requires appraisers, fiduciary agents, and off-chain data; points of centralization that create flaws in a system that claims to decentralize.
Regulatory side: while MiCA provides clarity in Europe, it limits innovation capacity for startups developing new DeFi and banking infrastructures. Focused on regulation and barriers to entry to slow the inevitable, European financial institutions are depriving themselves of the players who could help them regain leadership. And on a 10-15 year horizon, a systemic threat institutions must anticipate today: quantum computers will render RSA and AES—the two pillars of financial cryptography—obsolete. HSBC has already deployed post-quantum VPNs. Migration to PQC algorithms selected by NIST must start now, not in 2030. Here too, solutions exist and innovative startups are on the front line to provide them. However, the sector’s lack of open innovation risks turning against market makers like banks, states, and major European IT providers. The road to sovereignty is not won without a change in mindset.
What this means for the next 5 years
We are in the installation phase. The distinction between winners and losers will not be played out on balance sheet size. It will be played out on three simultaneous skills: regulatory compliance, technical security, and the financial logic of assets to be tokenized. For players thinking about finance, infrastructure, investment, or technology: the boundary between physical and digital assets is disappearing. A building, a bond, a commercial debt, a fund—all of this becomes programmable, fractionalizable, tradable 24/7, and usable as collateral in real-time.
McKinsey summarizes the situation with a formula I find hard not to see as definitive:
« Precision, not heft, is the great equalizer. In the age of AI, even smaller banks can capture disproportionate rewards by embedding precision into every dimension of strategy. »
The finance of tomorrow will not be bigger. It will be more precise, more connected, and more autonomous. The window to participate in the construction of this infrastructure, rather than suffering its disruption, is still open.
But it is closing.
Retrouver cet article en Français sur LINKEDIN : Cliquez ici
No responses yet