# Why Podcasts: The Expert Network Investors Always Wanted

> Top funds treat podcasts as a primary research input. 200M hours of expert conversation, often more candid than compliance-monitored calls, should be an input into your own investment research.


> **TL;DR** — Insight mining podcasts are becoming a critical input into the research process of fundamental investment managers. AI makes it possible to mine 200 million hours of expert conversation, aggregated into actionable insights on demand.

## Why podcasts and why now

Podcasts have exploded in popularity in recent years, but you may not have heard how elite investors and money managers are using them to mine for investment insights at scale.

Here is just one example out of thousands that we unearthed using Matterfact's Podcast tool. It is AI agent that sits on top of millions of podcasts and acts like a research assistant who can build high-quality, polished, and ready to use research reports.

Consider this: Between November 2025 and April 2026, CoreWeave ($CRWV) CEO Michael Intrator shared the same point across several podcast and CNBC appearances. **But each time, he added a little more detail.**

In November, he said the company had started the year with **80% of its backlog tied to a single customer**. By early January, that number had dropped, and no customer represented more than 30% of the book. Later in January, he put the figure at 32%. By April 10, he said it was under 35%, and publicly named **Meta, Anthropic, NVIDIA, and OpenAI as anchor customers**.

![Timeline of Michael Intrator's CoreWeave customer-concentration disclosures across podcast and CNBC appearances, November 2025 – April 2026](/assets/images/blog/timeline-podcast-insight.webp)

These granular details matter to investors. **Over 23% of the outstanding float has been sold short** (According to NASDAQ and CapIQ) and for short sellers, customer concentration presents a risk. It also matters if you were long NVDA and trying to understand how durable demand from neocloud players might really be.

What is interesting beyond the content is how and where the story unfolded.

Few investors could follow it fully because it did not show up in one filing or an earnings call soundbite. It came out over time, across several interviews, podcasts, and appearances that most investors would simply not have the time to listen to or read.

The funds that picked up on it did not necessarily have access to better people via an expert network, but they did have a better process for listening. At scale.

```request-access
variant: inline
heading: Search podcasts before your next expert call.
buttonText: Request access
```

Until now, podcasts were treated as entertainment, and optional listening for money managers. Interesting, sometimes useful, but not central to the research process at any serious hedge fund. Well, that is now changing and mining for insights from podcasts at scale may become as ubiquitous as using expert networks. **Some of the most candid, specific, and forward-looking commentary in the market now shows up in long-form audio before it appears anywhere else.**

The problem is, if you are not systematically data mining that resource, **there is a good chance you are missing critical information and competing against people who are.**

## How podcasts went from entertainment to investment tool

First, the quality of the format improved. Podcasters are "monetizing" this content and are after higher quality, longer and more in-depth interviews that attract more views and sponsorships & ad revenues. The guests are incentivized to provide quality content for a different reason. For many executives, founders, and specialists, podcast appearances are now part of how they build their own reputation and personal brand. They prepare diligently, explain things clearly and in detail, and they bring specifics to back up their claims and appear highly competent. And (this one is critical) because the format is longer and less formal than an expert network call, **they often say more than they would in traditional investor communication channels** or on a one-to-one call with an investor.

Second, the quality of guests is extremely high. If you scroll through investing podcasts, for instance, you will see a mix of practitioners, operators, subject matter experts (SMEs), and hedge fund managers that you would have never seen even five years ago. They were known for keeping strategy close to the vest. Many of these people are difficult to access directly, but surprisingly willing to spend an hour speaking in public on a pod. This is not only true of investing but every other industry as well. Noam Brown (From OpenAI) went on the Redpoint podcast to explain reasoning as a new paradigm of scaling. World-renown economists are commenting on the war in Iran and leading experts in cloud computing are discussing market-moving innovations.

Third, the specialist layer got much deeper. If you are looking to unearth insights that are not common knowledge on Wall Street, **the real opportunity is often in niche podcasts focused on specific industries** like semiconductors, freight, biotech, industrial automation, healthcare IT, energy infrastructure, specialty chemicals, and dozens of other verticals. That is where you start hearing informed people talk about things that truly matter for companies and whole sectors.

To ground that in data: across the episodes we index, Information Technology and Real Estate dominate the conversation, but every GICS sector has meaningful coverage — hover any bar to see the exact share.

```chart
podcast-episodes-by-gics
```

So the question is no longer whether useful signals exist in podcasts, it clearly does. The issue is volume and the sheer scale of content out there.

## The scale problem with using podcasts for investment research

There is an enormous amount of audio in the world. On top of the millions of podcasts out there with hundreds of millions of hours of recorded conversation, thousands of new episodes are published every day.

**No analyst (or even a team of analysts) can keep up with that volume manually.** The bottleneck is the ability to search, filter, compare, and retrieve the insights that matter to you and your org and it's akin to finding a needle in a million digital haystacks.

Thankfully we have AI. Now, instead of listening at 2x speed for hours in the hope of finding one useful detail on a name in their portfolio, **an analyst can simply ask a question and query the full corpus.** It is an AI agent that can pull the relevant data, compare discussions across hundreds of sources, and let the analyst spend their time doing what actually matters: interpreting the signal.

We are seeing managers use this data in two distinct ways. Idea generation and building a better investment thesis. We are sure more exist, like red flag detection and risk management, but will focus on the two common use cases for now.

## Use case one: idea generation

Every investor works with some version of an investable universe and a certain concentration threshold. They wake up every day thinking about the next stock to buy, either to replace a name that is on its way out, or to get exposure to a new theme.

Podcasts can be a surprisingly good resource for coming up with new investment ideas.

There are "whispers" about NVIDIA's AI strategy and the cost of new GPU clusters. There is talk about how legacy tech companies are becoming exceptionally cheap relative to the new commers. A former product lead at a big AI company explained why a platform shift will likely pressure margins across an entire category. One podcast is interesting but millions begin to formulate a theme, and an investment opportunity.

These things happen constantly on podcasts and most of the new ideas go unnoticed by wall street, save a few savvy investors who know how to mine these insights.

Here is another example. On the All-In podcast in March 2026, Perplexity CEO Aravind Srinivas cited OpenAI CFO Sarah Fryer on a point that has implications for how investors think about the economics of AI in general: **the cost of a million output tokens had fallen from roughly $32 at the GPT-3 launch to roughly $0.09 today.** That is a dramatic decline especially given that much of the reduction came from **software** rather than hardware alone.

Just imagine the questions you could ask of this vast body of data to help you identify your next idea. The question itself becomes your value. What would you ask? Send us your question and we will (privately) reply with a full report.

```request-access
heading: What if you could scan 100s of episodes for your question?
description: Send us your question — we'll reply privately with a full report.
buttonText: Send us a question
```

## Use case two: thesis enrichment

Idea generation is valuable, but building a bullet-proof thesis may be the more important use case, especially for analysts pitching a new idea to the PM.

Analysts already know how to build a DCF model and form a view and know exactly how to present it to their portfolio manager. Where things get harder is substantiating that view with real world evidence or even having someone poke holes in it before they make the pitch to the PM.

Expert networks were one of the main ways funds tried to bridge that gap; and they still have value. But they are not without limitations.

Experts networks are not cheap and experts are warned about saying too much. On a formal, compliance-approved call, people tend to be careful. On a podcast, especially a friendly one, they are often looser, more opinionated, and more willing to explain what they actually think is going on. Also, investors will often need to speak to multiple experts about the same theme - this takes time.

Mining podcasts for alpha is much cheaper and provides instant value, drawing from a large pool of experts. **It can give you a substantiated bull or bear case on almost any public company.**

Take the current debate around GPU depreciation and asset life in AI infrastructure. One version of the bearish case says that GPUs are likely to become obsolete quickly, which means today's aggressive capex could age badly and lead to painful write-downs later. That is an ongoing debate with experts arguing both sides, and investors need to consider both sides of it when crafting their thesis.

```request-access
variant: inline
heading: See what CIOs and economists are saying on podcasts.
buttonText: Request access
```

## Podcasts are still underused as an investment research tool

What makes this especially interesting is that the market has not taken advantage of all the insights hidden in podcasts yet.

A handful of sophisticated firms have, but most small and mid-sized funds still do not have a proper workflow for it. They are not systematically indexing, querying, comparing, and extracting signal from the medium and that creates an opportunity.

We have seen something similar with credit card data and other alternative data sets. A new source of alpha rewards investors who get serious about it before everyone else does. First it was satellite data, then transaction data, app data, web traffic, or earnings call sentiment parsing. **Today, one of the most underappreciated sources of differentiated public signal is long-form spoken content.**

Because it is too large to manage manually and still not widely integrated into research workflows. That is exactly the kind of inefficiency that tends to persist and offer ample alpha for early adopters.

```request-access
heading: Put the podcast layer behind your next thesis.
description: Matterfact is deployed with select institutional partners. Request access to run it on your own coverage.
buttonText: Request access
```