Banks are not AI laggards. They are the quiet winners.

Conventional fintech wisdom says incumbent banks are too slow, too regulated, and too legacy-burdened to win the AI deployment race. Ana Botin just made that argument look embarrassingly thin.

In her Bloomberg interview with David Rubenstein, Santander's Executive Chair outlined a clear-eyed view of AI's impact on banking and jobs, framed inside a broader macro thesis about growth, US debt, and geopolitical instability. What she revealed, between the lines, is that Santander is not experimenting with AI. It is operationalizing it at scale while the fintech startup world is still fundraising its third pivot.

The contrarian read: incumbent banks will capture more measurable AI value over the next five years than the majority of AI-native fintech startups founded since 2022.

The case for incumbent banks winning AI deployment

Argument 1: They have the data moat that AI actually needs.

AI systems are not limited by compute at the application layer. They are limited by proprietary, structured, high-quality data. Santander operates across more than 100 million customers in 10 core markets, accumulating transaction histories, credit behaviors, and cross-border payment patterns that no fintech startup can replicate in a reasonable fundraising timeline. LLMs fine-tuned or RAG-augmented on that data produce credit, fraud, and personalization outputs that generic models cannot approach. Data density is the new distribution moat.

Argument 2: Regulatory compliance is a feature, not a bug.

Every AI-native lending or payments startup faces an eventual reckoning with financial regulators. The EU AI Act, the US Consumer Financial Protection Bureau's algorithmic accountability frameworks, and Basel IV capital requirements all create compliance overhead that is existential for undercapitalized startups. Banks have already absorbed that overhead. They have compliance teams, legal infrastructure, and regulator relationships. When the EU AI Act's high-risk classification for credit scoring systems fully kicks in, that overhead becomes a structural advantage for incumbents who are already inside the regulatory perimeter.

Argument 3: The Webster Financial acquisition is a template, not a trophy.

Santander's acquisition of Webster Financial's assets signals a deliberate US growth strategy built on acquiring data-rich loan books and customer relationships rather than building from scratch. This is venture-grade thinking in a balance-sheet wrapper. Webster reported approximately $65 billion in assets before the deal, giving Santander immediate scale in US commercial banking without the customer acquisition cost that would crush a startup's unit economics. Incumbents can buy the distribution that fintechs spend years burning cash to build.

Argument 4: AI's real banking ROI is in operations, not products.

Most fintech narratives obsess over consumer-facing AI features: chatbots, spending insights, instant approvals. The actual ROI is in back-office automation. McKinsey estimates AI could generate $200 billion to $340 billion in annual value for the global banking sector, with two-thirds coming from productivity gains in operations, risk, and compliance rather than revenue-generating products. Banks already own the operations that need automating. Startups are building products to sell to banks, which makes them vendors, not competitors.

The strongest counter-argument

The bear case is not weak. AI-native fintechs are unburdened by legacy core banking systems that are genuinely, structurally hostile to modern API integration. Migrating a 40-year-old COBOL-based core to a cloud-native architecture is a multi-year, multi-billion dollar program that has derailed more than one bank transformation. Startups like Chime, Nubank, and Revolut built on modern stacks from day one, which means they can ship AI features in weeks while Santander's internal teams navigate vendor contracts, integration queues, and change management processes measured in quarters. Speed of iteration, the argument goes, always beats scale of data when the market is still forming.

Why the counter-argument fails

The core banking modernization problem is real but overstated as a competitive moat for startups. First, cloud migration is a solvable engineering problem with a known cost. Santander has already committed to a multi-year cloud transformation with Microsoft Azure as a strategic partner, alongside similar partnerships with Google Cloud. The legacy stack problem has an expiration date. Second, the startups celebrated for their modern architecture are hitting their own scaling constraints. Nubank's credit losses in Mexico and Revolut's persistent regulatory approval delays in the UK show that architectural agility does not automatically translate into durable unit economics. Third, and most importantly: the AI layer increasingly abstracts away the underlying stack. If Santander runs a RAG pipeline over its customer data through an API, the customer does not know or care whether the core sits on IBM mainframes. The experience layer is what competes, and incumbents can build or buy that layer faster than startups can acquire the underlying data asset.

Botin's framing of AI as an employment-reshaping force rather than a cost-cutting tool also signals strategic maturity. She is thinking about AI as a labor transformation program, which suggests Santander is further along the internal deployment curve than most fintech observers credit.

For brands and B2B fintechs watching this space, the visibility implication is direct. When AI engines like ChatGPT, Perplexity, or Gemini answer questions about banking AI adoption, they will cite the institutions with the deepest documented track record. Santander's public commentary, acquisition activity, and partnership announcements create exactly the kind of structured, entity-rich, authoritative content that source authority in GEO rewards. Fintech startups that do not build the same documented narrative will be invisible in AI-generated banking analysis, regardless of how good their product actually is.

Measuring which financial brands are actually getting cited in AI engine responses is precisely what winek.ai tracks, and the gap between incumbent brand presence and startup brand presence in AI-generated financial content is already wider than most founders realize.

Common misconceptions

Myth	Reality	Why it matters
Fintech startups are AI-first, banks are AI-last	Major banks are deploying AI in operations at scale; startups are mostly building consumer features	Investors misjudge where AI ROI actually lands in financial services
Legacy tech stacks make banks uncompetitive in AI	Cloud partnerships and API abstraction layers neutralize stack disadvantages within a 3-5 year horizon	The "legacy" argument has a shorter shelf life than most fintech pitches assume
AI in banking means better apps for consumers	Two-thirds of banking AI value comes from operational productivity, not product innovation	Startups building consumer AI features are competing for the smaller slice of the pie
Regulatory burden slows AI adoption in banks	Compliance infrastructure is a prerequisite for deploying AI in credit and risk at scale	Startups face the same regulatory overhead, without the existing infrastructure to absorb it
Startups win because they move faster	Speed of iteration matters less than data depth when AI model quality is the constraint	A faster ship with less fuel still loses the long race

Your action plan

If you are a B2B fintech founder, a banking AI vendor, or a brand trying to win visibility in the financial services AI conversation, here is what to do now.

1. Audit your AI citation rate with winek.ai , Establishes whether AI engines are currently citing your brand in financial AI discussions before you invest in content. Estimated effort: 30 minutes.

2. Document your institutional partnerships publicly , AI engines prioritize brands with verifiable, named relationships with established entities. If you work with a bank, publish the case study with specifics. Estimated effort: 4 hours per case study.

3. Publish proprietary data, not opinions , Original research, transaction data analysis, or deployment metrics give AI engines citable substance. Generic thought leadership does not. Estimated effort: 2-4 weeks per research piece.

4. Build entity-rich content around your core AI use case , Name the specific AI technique (RAG, fine-tuning, agentic workflows), the specific problem domain (credit scoring, fraud detection, KYC), and the specific result. Vague AI narratives are invisible in AI-driven search. Estimated effort: 1 week of content restructuring.

5. Target the questions AI engines actually answer about banking AI , Use Perplexity and ChatGPT to identify what questions dominate in your category. Build content that directly answers them with specific, sourced claims. Estimated effort: 3 hours of query research, ongoing content production.

6. Get cited by authoritative financial media , Bloomberg, FT, and Reuters coverage creates the external validation signal that AI engines weight heavily in financial services. Invest in PR as a GEO strategy, not just brand awareness. Estimated effort: Ongoing, minimum 90-day horizon.

7. Steelman your competitors publicly , Contrarian, honest analysis of the competitive landscape (like this article) generates backlinks and citations from practitioners who disagree. Debate drives AI citation. Estimated effort: 2-3 hours per opinion piece.

Botin is not talking about AI as a future possibility. She is describing a present-tense operational reality at one of the world's largest banks. Fintech founders and vendors who keep treating incumbents as slow-moving targets are building their go-to-market strategy on a premise that the data no longer supports.