a16z Podcast: Inside the $13T Mortgage Machine

Date: September 11, 2025
Host: Andreessen Horowitz (A16z)
Guests:

• Angela Strange, a16z general partner (B)
• Tim Mayopoulos, former CEO of Fannie Mae & president of Blend (A)
• Mike Yu, co-founder of Vesta (D)
• Andrew Wang, co-founder of Valon (C)

Episode Overview

This episode of the a16z Podcast dives deep into the $13 trillion U.S. mortgage industry—the largest class of consumer debt in America. The conversation dissects why mortgage technology lags behind, the immense regulatory and structural barriers to innovation, and how fintech entrepreneurs are rebuilding the system from scratch. With leaders from Blend, Vesta, and Valon, the panel also explores how AI could reshape the industry, reduce costs, improve transparency, and transform the home buying journey.

Key Discussion Points & Insights

1. Why is Mortgage Tech Slow to Change?

• Structural Challenges to Innovation:
  • Deep government involvement and guarantees mean high regulation and standardization.
  • Quote:

    “There isn’t any big thriving private mortgage credit market at scale where there’s a lot of new or innovative products...all of this is really backed by the government.”
    – Tim Mayopoulos (02:23)

  • Standardization:
    Mortgage products are all uniform to enable massive global capital flow into U.S. housing.

    “If you try to even change one word of the loan documents...you’re not going to get the loan.”
    – Tim Mayopoulos (03:28)

  • Emotional & Infrequent Nature of the Transaction:
    • Home buying is rare, highly emotional, and Americans want trusted human advisors rather than just speed/convenience.

    “It’s a really highly personal experience...tied up in deeply-seated emotions about home and family.”
    – Tim Mayopoulos (04:38)

2. The Impact of Outdated Infrastructure

• On Borrowers:

  • Fragmented, inefficient systems increase origination costs, directly raising mortgage rates and making homeownership less affordable.
  • Mortgage assistance is stressful and opaque, like “sending information into the void.”

    “Just imagine...you apply for mortgage assistance, which is...a pretty time intensive process...a little bit like sending a lot of information into the void.”
    – Andrew Wang (07:50)

  • Better software could make the process transparent “like a pizza tracker.”

    “If all of this was more efficient, then you can kind of see every part of the process like a pizza tracker.”
    – Andrew Wang (09:20)

• On Lenders and the System:

  • Manual labor and opaque/fragmented data leads to high expense, poor customer experience, and hinders capital market innovation.
  • Data transparency issues create regulatory overhead and block new product development.

    “The lack of transparency...causes a lot of bigger downstream problems as well.”
    – Mike Yu (13:00)

3. Why Build New Cores—Not Just ‘Wrappers’

• Loan Origination Systems (LOS):

  • Incumbent systems are decades old; manual and monolithic; barely fit modern lenders’ strategies.
  • Lenders want composable, best-in-class, open-data architecture.
  • Building a new LOS allows lenders to truly reimagine workflows and reduce costs.

    “Mortgage lenders that are independent mortgage banks don’t really make money per loan on average today...cost to originate continues to balloon.”
    – Mike Yu (14:05)

  • Wrapping old cores doesn’t enable the operational transformation needed.

    “If we had tried to build something as a wedge around the whale that is Encompass, I think we...wouldn’t have been able to drive the fundamental change.”
    – Mike Yu (29:06)

• Servicing Side:

  • Andrew Wang describes servicing as a web of payments, collections, compliance, and customer experience that’s historically a ‘clusterfuck’—hundreds of systems at big banks.
  • Attempting piecemeal fixes doesn’t work—you need to rebuild the core to solve real problems and drive drastic profitability gains.

    “[It’s] this like Frankenstein of all these myriads of systems ... a clusterfuck.” – Andrew Wang (18:39)

  • Entered servicing “the hard way”—licensing in all states, building a regulated entity, then scaling with modern software.

4. Modernization Pathways: Fannie Mae, Data & Industry Evolution

• Fannie Mae's Approach:
  • Focused on enabling third-party tech providers.
  • Automation, APIs, digital data consumption, innovations like Day One Certainty all aim to drive digital adoption without increasing risk.

    “Virtually everything that relates to a mortgage loan is just a piece of digital data.”
    – Tim Mayopoulos (22:52)

  • Real bank data could replace old proxies like credit scores and appraisal methods, expanding access and reducing costs.

    “We should be consuming the real data...and be able to underwrite people more effectively.” – Tim Mayopoulos (25:25)

5. The Role of Artificial Intelligence (AI)

• Current Applications:

  • AI excels at reading, extracting, and processing mortgage documents—has finally made ‘document ingestion’ reliable at scale.

    “Reading things like purchase contracts or title insurance policies...are now frankly really simple...”
    – Mike Yu (35:49)

  • Valon uses LLMs for summarization, voice agents, and personalized workflow agents, moving toward task-specific AI that can “clone your best people.”

    “If people put their entire business process on our workflow engine, they then are able to build automatic AI agents for that specific task...”
    – Andrew Wang (38:20)

• Future Potential:

  • AI likely to subsume regulatory interpretation/checks, making compliance automated and reducing errors/costs for lenders and consumers.

    “Virtually everything relating to the regulatory aspects of mortgage will end up being subsumed by AI.”
    – Tim Mayopoulos (41:16)

  • Most value will accrue to users of AI, not model builders.

    “Most of the value in this AI revolution is going to accrue to the people who use the AI.”
    – Mike Yu (44:20)

• Guardrails & Risk:

  • Highly regulated nature of the industry means AI solutions must be explainable and auditable—FICO score example illustrates why simple models remain due to explainability.

    “You need explainability...because anything but that...when I get in front of the regulators...I won’t have a really great answer.”
    – Andrew Wang (46:17)

Notable Quotes & Memorable Moments

• Mortgage, not pizza—but maybe it should be!

  “We are going to title this podcast: Mortgage should be more like Pizza.”
  – Angela Strange (10:59)

• On opportunity and patience in mortgage tech:

  “There’s a class of problems out there where there’s this truly entrenched complexity that has suppressed competition long enough that there’s this astronomically large opportunity. But you have to be patient, you have to be relentless...”
  – Andrew Wang (20:03)

• The drive toward AI adoption:

  “AI is not your competitor, it’s your competitor that adopts AI before you.”
  – Angela Strange (39:30, referencing Jensen Huang)

Important Segments & Timestamps

• [02:23–07:20] Tim Mayopoulos on barriers to innovation in mortgage tech
• [07:50–10:59] Andrew Wang on inefficiencies for borrowers and the ‘pizza tracker’ analogy
• [13:05–16:18] Mike Yu on launching a new LOS and the demand for core replacement
• [16:37–19:49] Andrew Wang on the challenge of modernizing servicing, why few try
• [22:24–25:15] Tim on Fannie Mae’s approach to fostering industry innovation and real data underwriting
• [28:59–33:04] Mike Yu on rebuilding from scratch vs. wrapping old cores
• [35:49–39:43] Mike and Andrew on immediate & future uses of AI in origination and servicing
• [40:41–43:02] Tim on AI for regulatory logic and reducing human error
• [44:20–47:29] Mike and Andrew’s advice for lenders embracing AI
• [47:34–53:50] Closing visions — utopian and dystopian — for the mortgage ecosystem’s next decade

Looking Forward: Visions of the Mortgage System

• Mike Yu: Near-instant, automated, “one-touch” experiences; compliance and capital structures could evolve; back-office fulfillment will be invisible to consumers. (47:34)
• Andrew Wang: Anticipates a data-rich, AI-enabled world—possible “dystopian” levels of consumer surveillance but hyper-efficient transactions; capital markets could markedly shift as inefficiencies disappear. (48:23)
• Tim Mayopoulos: Optimistic about AI and innovation but urges regulators/GSEs to embrace and mandate change, not protect incumbents—true innovation requires policy adaptation. (51:01)

Summary Takeaway

The U.S. mortgage industry’s complexity, regulations, and legacy tech have kept innovation at bay. But determined fintech entrepreneurs are tearing apart the status quo—rebuilding core systems for origination and servicing, and unleashing the power of modern APIs and AI. As technology matures and policy adapts, the promise is both a more affordable, transparent process for consumers and a radically transformed financial system. The biggest winners? Those who embrace and operationalize technology—not just those who build it.

For industry leaders, regulators, and homeowners alike, this episode offers a roadmap to where the $13T mortgage machine is heading—and why the smartest players aren’t waiting for incremental change but instead are building for a future powered by open data, AI, and bold disruption.