Abstract
For fifty years, $400 trillion in professionally managed assets sat untouched by external technology. The most analytically demanding investment workflows required a combination of PhD-level quant finance and advanced engineering that no outside vendor could sustain commercially. Wall Street institutions captured those professionals, built captive systems, and kept everything proprietary. AI is the first force capable of breaking that lock, and this talk is a practitioner's honest account of building within that moment.
We examine three realities most conference talks sanitize.
- First: why the opportunity is structurally open for the first time, and why domain expertise practitioners already carry is now their most defensible asset, capable of powering what he calls the billion-dollar solo enterprise.
- Second: why the pre-AI startup playbook on talent, capital, and org design is not just outdated but actively harmful when applied uncritically today.
- Third, and most uncomfortable: AI creates billion-dollar companies overnight and destroys them just as fast. The next Yahoo will be the firm that mistakes capability advantage for a structural moat.
What You Will Leave With
- Why the $400T asset management market is structurally open to disruption for the first time, and what broke the lock
- How deep domain expertise becomes a defensible moat in AI-native company building
- Where pre-AI startup logic on talent, capital, and org design becomes a liability in the post-AI era
- Failure patterns of AI companies that built on capability rather than domain depth, and how to avoid them
- Four survival principles: data-layer defensibility, human-in-the-loop as product, redundancy as revenue strategy, and infrastructure first
Interview
The asset and wealth management industry sits on $345 trillion in investable wealth, generates ~$3 trillion in annual revenue, and spends over $300B a year on technology — yet ~70% of that spend goes to keeping legacy systems alive. This talk is a five-part reality check on what it takes to build AI-native fintech from the ground up to serve this industry: why the opportunity is unprecedented, why deep domain expertise — not technology alone — decides who wins, how AI enables hyper-customization at scale for the first time ever, how to architect for survival when foundation models change leaders every few months and go offline without warning, and what the economics of a $1B solo enterprise actually look like. For senior developers, the core question is architectural: the financial institutions you work with or inside are trapped by the very systems you maintain. Understanding the structural forces — and the sharp contrast between what incumbents can ship versus what AI-native startups can — will reshape how you think about what to build next.
Inside financial corporations: 74% can't scale AI past pilot. 63% don't have AI-ready data. Deployments average 14 months. 41% of young employees actively resist AI adoption. The machinery is stuck.
Outside — in AI-native fintech startups: deployment in weeks, not years. $1M+ revenue per employee versus $200–500K. 21% revenue growth versus 6% for incumbents. Yet fintech has penetrated only 3% of global financial services revenue. 97% of the market is still ahead.
The contrast has never been sharper. If you're building inside a financial institution, understanding why the wall exists is the first step to breaking through it. If you're building outside, understanding the domain is your only durable moat. Either way, 2026 is the year this divergence becomes irreversible.
Inside corporations, five headwinds dominate: the AI ownership debate (who controls the models and their outputs), data ownership paralysis (63% lack AI-ready governance), bureaucratic inertia that turns a 2-month build into a 14-month procurement cycle, fear-driven resistance from teams whose roles feel threatened, and risk aversion that kills projects after proof-of-concept. These aren't technology problems — they're institutional ones. The code works. The organization doesn't.
For startup builders, the challenge flips: you have speed and freedom, but finance demands complex data logic, sophisticated math, regulated workflows, and deep institutional trust. Bloomberg survived every tech wave since 1981 not because data was scarce — because the workflow is irreplaceable. Aladdin charges $5–20M a year not for features — for lock-in. If you build without domain depth, the next model upgrade replaces you. Domain is not background — it's the product.
Audit your AI architecture for single points of failure. I show real outage data — ChatGPT down 12+ hours, Claude down 3 hours, Cloudflare cascading across every AI platform. In finance, one hour offline is a fiduciary breach. Model-neutral, FM-diversified architecture with on-prem fallback isn't a design preference — it's a survival requirement. I'll show the exact pattern: LLM layer (model-neutral) → infrastructure layer (FM-diversified + on-prem) → your domain intelligence layer (proprietary, irreplaceable). That bottom layer is where your moat lives.
Practitioners talking to practitioners. No vendor keynotes dressed up as thought leadership. The audience has shipped systems at real scale, which means the questions cut straight to substance. I can present honest numbers — including failures and near-misses — and the room engages with both. That's rare.
Topics
QCon London 2026 is a three day conference for senior software engineers, architects and team leads. An international program committee of working engineers selects every session. Patterns and practices, not products and pitches.
From the same track
Monday 16 March
10:35 Mountbatten (6th Fl.) Session architecture How To Run on Three Clouds at Once, and When Not To Ross McFarlane, Kevin Holditch If you've made a bank transfer recently, there's a good chance that Form3 handled it. When we have a wobble, people notice. 11:45 Windsor (5th Fl.) Session AI/ML Your Multicloud Strategy Is a Product Problem - Treat It Like One Luis Henrique Albinati Junior, Surabhi Mahajan You can't really "opt out" of multicloud anymore. Between cloud concentration risk, SaaS sprawl, and increasing regulatory expectations, most enterprises end up operating across multiple clouds whether they planned to or not. The hard part isn't having two or three providers. 13:35 Rutherford (4th Fl.) Unconference Unconference: Tech of Finance Industry 14:45 Mountbatten (6th Fl.) Session Alpha Copilot The Realities of Building an AI Native Fintech Startup David Lin Founder and CEO @Linvest21, Previously CTO @JPMorgan For fifty years, $400 trillion in professionally managed assets sat untouched by external technology. The most analytically demanding investment workflows required a combination of PhD-level quant finance and advanced engineering that no outside vendor could sustain commercially. 15:55 Mountbatten (6th Fl.) Session Platform Engineering Move Fast, Don’t Break Trust: Shipping Constantly with Humans and Beyond Suhail Patel Senior Staff Engineer @Monzo Leading the Platform and Data Functions, Previously @Citymapper Moving fast only works if your systems are engineered with trust and verification in mind. In this talk, we showcase how teams can scale to the point where shipping hundreds of changes a day becomes boring, routine, and safe.