❓ Problem
Global podcast ad spend is on track to hit around $5 billion in 2026, up roughly 20% year on year, and it is growing faster than digital advertising as a whole. This is almost exactly the stage Facebook ads were at back in 2015: real money, real audiences, but the legacy agencies still would not touch it. Then they did, and the spend doubled every year after.
Here is the difference. Facebook eventually built a perfect auction market for its inventory, where every impression got priced live by advertisers bidding against each other. Podcasting never did. Pricing still runs on rate cards and relationships. A host reads out a CPM, an agency haggles, a spreadsheet gets emailed, and a slot gets booked weeks out. Host-read mid-rolls in the US typically go for $35-50 CPM, with premium finance and business shows commanding a 30-60% premium on top, but none of that is set by a live market. It is set by whoever picks up the phone.
The worst of it is the last-minute inventory. A slot that has not sold by the time an episode drops is gone for good, and unlike a billboard it cannot be resold tomorrow. Right now that unsold inventory just gets filled with house promos or given away, which for a show doing real numbers is money set on fire every single week. There is a live pool of demand and a live pool of perishable supply, and no exchange to clear them against each other. Here is the idea.
✅ Solution
A real-time bidding exchange for podcast ad slots, so unsold and last-minute inventory gets auctioned off instead of wasted.
- Creators list a slot with a floor price and a hard deadline. The mid-roll on Thursday's episode, floor $40 CPM, bidding closes when the episode publishes. Perishable inventory, priced like it.
- Advertisers bid against each other as the clock runs down. Live price discovery, the same mechanic Facebook built for display, pointed at a market that has never had it. A slot that would have gone unsold at any price now clears at whatever the market will actually bear.
- The winning ad gets generated in the host's real voice. Once a bid is locked, the brand shares its creative, the host approves it, and an AI model trained on that host's own back catalogue produces the read in their voice. This is the unlock. It means a slot can be sold and filled up until the last minute, with no scheduling call, no studio time, no waiting on the host to record.
The wedge is narrow on purpose: sell one host's perishable inventory that would otherwise expire. The exchange underneath, floors plus live bidding plus instant AI fulfilment, expands into the whole back half of the podcast ad market.
📊 Key Numbers
Market size
- Global podcast ad spend is heading for around $5 billion in 2026 and, on a low-double-digit growth rate, comfortably north of $8 billion by the end of the decade. That is the total pie an exchange sits on top of.
- Programmatic already accounts for roughly 22-29% of spend, so call it a ~$1.2-1.4 billion slice that is already comfortable transacting through automated pipes. Layer on the host-read remnant and last-minute inventory that currently sells for nothing, and the addressable gross flow for an exchange is realistically $1.5-2 billion of ad spend.
- The gap the exchange arbitrages is stark: programmatic podcast CPMs run $5-18 while host-read runs $35-72, a 4-5x spread. Every last-minute host-read slot that today drops to a remnant CPM (or zero) is inventory being sold at a fraction of its live-market value.
ARR potential
- An exchange monetises on take rate. At a standard ~15% take on $1.5 billion of gross spend transacting through the platform at maturity, that is a ~$225M revenue business. You do not need the whole market, you need the perishable tail of it.
- Bottom-up and early: route just $75M of gross ad spend through the exchange in the first couple of years at a 15% take and that is ~$11M in revenue, before any AI-fulfilment fee on top.
- Realistic path to $15-25M ARR within a few years as the default rail for last-minute podcast inventory, with a $100M+ ceiling if it becomes the auction layer the whole category eventually standardises on.
⏰ Why Now