Based on market data originally reported by Brander Group (June 2026 IPv4 Trends Report).

If you had asked most network operators in 2024 where the IPv4 market was headed, the answer would have sounded reasonable: downward, slowly. IPv6 penetration kept climbing, hyperscalers had stockpiled what they needed, and the wild pricing of 2021–2022 had given way to a calmer, buyer-friendly environment. IPv4 looked like an asset class settling into retirement.

The numbers from the first half of 2026 make that consensus hard to defend.

Start With the Routing Table Paradox

Here is the puzzle worth sitting with. Between January 2025 and January 2026, the global IPv4 BGP table expanded from roughly 996,000 routes to about 1.05 million—growth of around 54,000 prefixes, or 5.4%. By mid-June of this year, it had climbed another 11,500 routes, crossing 1.06 million advertised prefixes.

Meaningful growth, yes. But compare it against the volume of address space actually changing ownership: roughly 29.7 million IPv4 addresses transferred in the first half of 2026 alone, on top of 58.1 million across all of 2025. Ownership is churning at a scale that dwarfs what shows up as new announcements.

There are two ways to read that gap. If buyers were mostly financial players parking blocks in portfolios, transfers could spike without touching operational networks. The evidence points the other way: acquired space is being absorbed into networks that already exist—folded into established aggregates, placed behind customer-facing services, consumed inside existing infrastructure rather than announced as fresh prefixes. In other words, the addresses aren’t sitting in vaults. They’re going to work.

That matters for anyone watching supply. Address space that enters production rarely comes back to market on any short timeline. Heavy transaction volume, counterintuitively, can mean less future availability, not more—the market is burning through supply faster than sellers replenish it.

The Volume and Price Signals Agree

Zoom out to the raw transfer statistics. About 24 million addresses moved in the first five months of 2026—already more than half of 2025’s full-year total, and better than 28% ahead of the prior year’s pace at the same point. Extrapolated, 2026 lands near 59.4 million transferred addresses, edging past 2025’s record by roughly 2.2%. Measured against the five-year annual average of about 46 million, this year is running approximately 28% hot. Even a sluggish second half would leave 2026 well above historical norms.

Pricing tells the same story from the other direction. Large, clean blocks—/16s, /17s, /18s—that traded near $9 per address not long ago now fetch $13 to $16, a 20—30% move depending on registry, history, and reputation. Smaller prefixes, from /24 up through /19, have run harder still, commonly clearing between $18 and $24 per address as buyers pay up for space they can deploy immediately.

When volume and price rise together, the textbook interpretation is straightforward: demand is pressing against supply, and buyers are competing for the shrinking pool of well-maintained, readily usable space. Compounding this, APNIC’s annual BGP analysis shows roughly 84% of IPv4 routes are now /24, /23, or /22 announcements—the industry keeps slicing address space into smaller operational units, each carrying its own route objects, RPKI records, and filtering overhead. The scarce commodity is no longer just addresses; it’s clean, low-friction addresses.

Look at Who’s on the Recipient Lists

The character of the buyer pool may be the most telling shift of all. In earlier phases of the market, the biggest acquirers skewed toward hosting companies, regional ISPs, and investors betting on appreciation. This year’s recipient lists read differently: they’re dominated by operators of large production platforms.

Two May transactions illustrate the point. Vodafone took delivery of roughly 400,000 addresses—the month’s largest acquisition—with Salesforce close behind at just over 300,000. Neither profile fits speculation.

A carrier like Vodafone faces demand from every direction at once: subscriber additions, enterprise services, mobile buildout, IoT fleets, broadband expansion. IPv6 is genuinely advancing across telecom, but the operating reality remains dual-stack, which keeps IPv4 in the category of working infrastructure rather than legacy debt awaiting retirement.

Salesforce represents a different pattern: the modern SaaS company that has quietly become an infrastructure company. AI features, tenant environments, analytics pipelines, integrations, cloud workloads—each one terminates, somewhere, in networking resources. Firms building next-generation cloud and AI platforms are buying IPv4 even as they invest aggressively in what comes after it.

Two Demand Engines Still Warming Up

Both of those buyer stories connect to macro forces that most IPv4 forecasts haven’t fully priced in.

The first is AI infrastructure. Public attention fixates on GPUs, power draw, and headline-grabbing data center budgets, while the networking layer underneath gets little scrutiny. Yet every training cluster needs connectivity, every inference endpoint needs infrastructure, and every AI-powered product needs load balancers, security tooling, monitoring, and global distribution—a stack that still runs substantially on IPv4. The AI buildout is generating second-order address demand across the whole networking ecosystem.

The second is U.S. broadband funding. After years of administrative runway, BEAD dollars—more than $22 billion allocated nationally—are finally converting into real deployments. The causal chain is simple: new fiber creates new subscribers, new subscribers require new infrastructure, and new infrastructure consumes addresses. IPv6 will carry a growing share of these greenfield networks, but dual-stack remains the default architecture, which means BEAD-funded buildouts will pull IPv4 along with them.

Forward Indicators Point the Same Way

Transfer statistics are a rearview mirror; registry request queues are a windshield. At ARIN, transfer requests rose from 135 in April to 154 in May—a 14.1% month-over-month jump that also cleared the 2025 monthly average of roughly 148. A request doesn’t reach ARIN on a whim: by that stage, budgets are approved, network plans are drawn, and the business case is settled. Sustained request growth signals where organizational demand is heading, not where it has already been.

The Takeaway

Line up the indicators: annual transfer volume tracking toward a record above 59 million addresses; run rate roughly 28% over the long-term average; large-block prices up 20—30%; carriers and SaaS platforms—not speculators—anchoring the recipient lists; ARIN requests accelerating; AI and BEAD demand still early in their curves. Nothing in that list describes a market in structural decline.

The last two years of commentary dwelled on softening prices and easy inventory. The first half of 2026 suggests that was the intermission, not the ending. Scarcity didn’t leave the IPv4 market—attention did. The transfer data indicates attention is coming back.