Why Vertical AI in Financial & Professional Services Will Create $1 Trillion Enterprise Value

This article was originally published on Elsa Capital's website.

Introduction

For decades, technology waves like the Internet, mobile, and cloud have transformed industries from retail to media to transportation. Yet financial and professional services—representing over 20% of our GDP and $30 trillion in enterprise value—remain stuck with technologies from over 40 years ago. The tech stack of a finance professional is still primarily Microsoft Office, Bloomberg, and clunky in-house tools.

Despite advanced education, millions of finance professional spend most of their day on mundane, repetitive tasks. It's not the long hours that are the problem—it's what fills those hours: often low-value grunt work. Generative AI now creates the first real opportunity to change this reality.

I started Elsa Capital because I believe vertical GenAI applications in financial & professional services are going to create over $1 trillion in enterprise value. In addition to returns, this is also about creating a future where financial professionals find more meaning and joy in their work. 

A future of both profit and purpose.

A Personal Note 

I've spent the first half of my career in financial services and the second half in tech. Many assume I left finance because I hated it, seeking light in the exciting tech world. But that wasn't the case.

I left Fidelity for Stripe because I was curious about the inner workings of building and operating durable, high-growth companies. Even in tech, I gravitated toward financial services—having worked at a payments innovator and built SaaS product for the CFO org. Finance remains my root, where my career began and where many of my close friends and connections remain.

Why start a VC firm? Because I believe we're at an inflection point to see the largest transformation in financial & professional services in over 40 years—driven by Generative AI.

Financial Services: A Massive Sector Underserved by Technology

The financial and professional services sector represents ~25% of the global market cap through public companies alone—banks ($15.7T), insurance ($4.3T), investment firms ($5.5T), and professional services ($2.2T). Private partnerships like accounting firms, hedge funds, and law firms could add another $5T if they were public.

According to the Bureau of Economic Analysis, financial & professional services sectors contribute over 20% to U.S. GDP, double the Tech sector's 9%.

Yet this massive sector has seen minimal technological innovation for 40 years:

• Microsoft Excel was launched in 1985

• Bloomberg Terminal was built in 1982

• FactSet dated back to 1978

Previous technology waves transformed other sectors dramatically:

• Internet transformed Retail and Media (e.g., Amazon, Google, Facebook)

• Mobile revolutionized Consumer in terms of how we transport, eat, live, and entertain (e.g., Uber, Doordash, Airbnb, TikTok)

• SaaS shifted value from on-premises software to cloud (e.g., Salesforce, ServiceNow)

Throughout all these technology waves, financial services remained largely unchanged, continuing to operate on technology founded over 40 years ago, back in the Computer era.

Why Now: Generative AI is Going to Fundamentally Transform Financial & Professional Services (in a Bigger Way than Computers) 

About a year ago, generative AI emerged, giving us a glimpse of its potential to completely transform financial and professional services jobs, especially at junior levels and in roles containing large amounts of repetitive, mundane work.

 For the first time in 40 years, financial services firms started to care deeply about technology—specifically GenAI. And that level of attention is not just bottom-up but comes directly from the C-suite across CEO, COO and CTO. Why?

1.   Document-Heavy Workflows: Financial and professional services are filled with document/text-heavy processes that are ideal use cases for GenAI. One example: during my time in investment banking, I spent two months creating 200 company overview slides ("one-pagers") for potential acquisition targets. Over 90% of that process involved manual copy-paste, extracting information from documents, synthesizing it, and creating standardized outputs—precisely what GenAI excels at.

2.   Human Capital Intensity: Financial and professional services are fundamentally human-capital intensive. The largest cost item is compensation. Just three banks—JP Morgan, Bank of America, and Morgan Stanley—paid over $100B in compensation costs in 2024. For law firms, accounting firms, and consulting firms, ~75% of the cost structure is labor.

3.   Competitive Pressure: Financial and professional services operate in highly competitive environments (unfortunately no monopoly or oligopoly) with relatively thin margins. Hence, there's strong incentive to adopt AI for efficiency gains and real fear-of-missing-out (FOMO).

For a bank to improve profit, you can either bet on interest rates or market activities – which they can't control – or focus on basis points (bps) of margin improvement, which they can control. For private firms and partnerships (hedge funds, law firms), there's substantial incentive to reduce costs, so partners can pocket each additional dollar of profit in their own distributions. The founder of Millennium Management earned $3.8 billion in 2024, while Kirkland & Ellis partners collectively earned $5.3 billion in profit, with each equity partner averaging $9 million distribution. The incentives for adopting GenAI to reduce costs are evident.

 Beyond the business case, there's a human element that matters equally. 

Creating a More Fulfilling Future of Work

For decades, jobs in financial & professional services have been filled with manual, mundane tasks—whether for accountants, lawyers, investment bankers, compliance officers, insurance brokers or underwriters.

The Industrial Revolution provides a relevant analogy. Before the Industrial Revolution (beginning in the 1760s), 80-90% of people worked in agriculture, performing physically repetitive labor. Technological advancements reduced the need for manual farming labor, and agricultural employment dropped to 22% of the workforce by the 1840s. This shift freed people from physically intensive, repetitive jobs, allowing them to move to more enjoyable work in factories and later offices.

Today's knowledge workers in financial & professional services are in some ways similar to pre-industrial agriculture workers – performing mundane, low-level tasks, just on a cognitive level rather than a physical one.

Our hope is to reinvent the future of financial & professional services work – taking away the grunt work (where AI agents could do better) to make them more meaningful, creative, and joyful. 

The $1 Trillion Value Creation Opportunity 

I am not a traditional venture capitalist, and I did not grow up in the TAM world. As such, I've triangulated the value creation potential from three different angles. While no market sizing is perfect, this analysis gives me conviction that we are looking at a trillion-dollar ballpark for value creation opportunity for AI in financial & professional services. 

Approach #1: Bottom-Up Labor Cost Analysis

Financial and professional services are fundamentally human-capital intensive. And the exciting part of GenAI applications (or agents) is their potential to augment and eventually replace some of these roles.

In the U.S., there are 11.8 million financial and professional service (PFS) roles, with a weighted average base salary of $91,393 (source: Elsa Capital analysis, based on the Bureau of Labor Statistics employment data).

Adding in bonuses (assume 25% of base) and benefits (assume 25% of base), the total labor costs for FPS jobs in the U.S. is around $1.6 trillion. Conservatively estimating global costs at twice the U.S. figure brings us to $3.24 trillion globally.

If vertical AI applications capture only 10% of these labor costs (a conservative estimate given AI tools can already outperform human banking or hedge fund analysts in specific areas), that represents a $324 billion ARR opportunity. 

Applying a 5x EV/Sales multiple (conservative relative to the 7x average for SaaS companies and the 10x for financial data companies), we arrive at $1.6 trillion in enterprise value creation.

Source: Elsa Capital analysis

Approach #2: Top-down Labor Costs Analysis 

Another way to size the labor costs is take a top-down approach estimating labor costs as a percentage of revenue or market cap. For the world's largest banks like JP Morgan (8%), Morgan Stanley (14%), and Bank of America (13%), labor costs represent on average 11% of market cap.

Extending this analysis across the entire financial services and insurance industry (given similar business models) and also adding in private firms like accounting and law firms, this approach yields total labor costs on financial & professional services of $3.4 trillion, aligning with Approach #1.

Source: Elsa Capital analysis

Approach #3: Market Comparable (Comps) Analysis

Today's financial and professional data and software providers command $520 billion in enterprise value. These companies primarily aggregate structured, tabular data, representing less than 10% of what professionals in these fields actually do.

 The remaining 90% of professional work involves processing texts, analyzing thousands of pages of financial, legal, or tax documents.

 If addressing only the tabular data portion has already created over $500B in enterprise value, generative AI that automates intellectual processing of unstructured documents should reasonably create multiple times of that value, i.e. in trillions. 

Source: Elsa Capital analysis, market data as of 3/20/25

Final Note

We are just at day 1 of massive value creation in vertical AI for financial & professional services in the coming decades. Although there are still many challenges ahead—from AI-specific issues like hallucinations and the chasm between pilot and multi-year contracts, to general startup challenges —we are going to see enormous opportunities for value creation and investment returns.

We believe now and the next 20 years is the unique time to seize this opportunity. The technical capabilities have matured, the market is open (in fact there is strong market pull), and the ROI case is compelling for financial and professional services firms to adopt AI.

We are fortunate to already have the opportunity to partner with amazing startups building AI agents or co-pilots for specific financial and professional service roles. These include Portrait Analytics (AI for public market investors), Pearson Labs (AI for law firms), Revi (AI for M&A deal prospecting), Dealops (AI for in-house deal pricing), Bayesline (AI for financial and risk analytics), Yellowpad (AI for in-house counsels), and Lucite (AI for insurance brokers).

One piece of advice often given to startup founders is to solve a problem you've personally experienced. This has a twofold advantage – one is that founders already come with deep domain expertise and unique insight, and the other is that founders genuinely care about the problem (because the journey of entrepreneurship is so hard that emotional connection matters).

As the founder of an investment firm, I feel fortunate to have the opportunity to invest in products that the younger me would have loved to use, and tools that I want many of my financial and professional services friends and connections to adopt to find more joy and meaning in their work.

About the author

Sarah Fu is an experienced operator, investor, and founder with 15 years of experience at the intersection of finance and technology. At Stripe, she was the strategic CFO for the flagship product, advised the C-suite on strategy, and led billion-dollar fundraising efforts. As an investor, Sarah has overseen $8B in public investments on behalf of Fidelity and is an active angel investor and advisor in GenAI, Fintech, and B2B startups. She holds an MBA from Stanford and BS in Bioengineering and Finance from the University of Pennsylvania.