For two years, the IPv4 market collectively decided that cheaper meant fixed. Prices dropped, buyers got comfortable, LinkedIn got flooded with IPv6 hot takes, and somewhere a broker in a WeWork told a client to “just wait it out for better prices.” Great advice. Aged beautifully.

24 million addresses have already transferred in 2026—annualized is on track for 71 million. That’s 23% above 2025 and 50% above the five-year average. Apparently the wait-it-out crowd is still waiting while everyone else moves inventory at a pace the market hasn’t seen in years.

Pricing on /16s and larger opened the year at around $9 per IP. It’s now $12–$15. That’s a 33–55% move in five months. Move in only a few months. The buyers who listened to the “market is structurally broken” thesis are now paying a third more than the buyers who didn’t. The market has opinions about bad analysis, and it expresses them through your invoice.

Watch the buyers, not the narrative. Hostinger pulled 393k addresses last month. BytePlus—ByteDance’s cloud and AI infrastructure arm—absorbed 360k. These aren’t legacy telcos filling compliance spreadsheets. These are companies aggressively building infrastructure for actual growth, and they’re doing it with IPv4 because that’s what the internet still runs on, regardless of what the conference panel said.

BytePlus deserves a moment. This is the network substrate underneath one of the most aggressive AI and content machines on the planet. When that organization quietly acquires 360k addresses, it’s not legacy behavior. It’s the infrastructure layer that all the breathless AI coverage doesn’t mention because it’s boring and physical and doesn’t fit on a slide deck. The models need networks. The networks need IPs. The IPs are finite. This is not complicated.

Now, IPv6. The perennial “this is finally the year” asset of the networking world. IPv6 has been almost-ready-to-matter since approximately the second Bush administration. Sixteen years of imminent disruption, and IPv4 is still handling the routing, the security policies, the customer environments, and the operational muscle memory of every network team that has better things to do than rearchitect working infrastructure on a timeline set by an IETF document. IPv6 will matter. It has not mattered enough to move this market yet, and the people who bet on it doing so in the last 24 months paid for that optimism in real dollars.

BEAD is the slow-moving train that the market keeps forgetting is still on the track. Years of grant applications, revised maps, appeals, more revised maps, federal comment periods, and the general bureaucratic performance art that precedes any government infrastructure program. That process is now becoming construction. Fiber goes in the ground. Customers get connected. ISPs need address space. They will arrive in a market already being squeezed by AI infrastructure companies and hyperscalers, holding freshly approved BEAD grants and a sudden, urgent need for IPv4. The timing is genuinely not great for buyers.

April’s ARIN transfer request count came in at 135, which is below the 2025 monthly average of 150. Certain observers treated this as vindication. It isn’t. Transfer requests and transfer volume are not the same metric, and conflating them is the analytical equivalent of judging a restaurant’s revenue by counting reservations. Fewer transactions moving larger blocks is what serious infrastructure acquisition looks like. One soft month in a count metric doesn’t undo five months of price increases and accelerating volume.

The honest summary: the pool didn’t get bigger when prices dropped. Sellers got anxious, buyers got patient, sentiment curdled, and the market spent two years mistaking a pricing correction for a structural shift. It wasn’t. The correction ended. The buyers noticed before the sellers did, which is why prices are already up 33—55% and large clean blocks are getting genuinely hard to source.

By end of year, /16s and larger are likely sitting at a minimum of $18–$22 per IP. The days of waiting for distressed sellers are gone. It’s time to face the new reality.

Scarcity was always there. The market just decided, for a couple of years, that acknowledging it was optional. That option has expired.