RIPE NCC Annual Report 2025: What an LIR Should Read First

Picture the day your finance lead drops the RIPE NCC renewal on your desk and asks one question: do we just pay it, or is there something in here we should act on first? That decision lands on someone in every member organisation this spring, and most of them get it wrong.

The RIPE NCC published two documents on 17 April 2026: the Annual Report 2025 and the Financial Report 2025. They are supporting material for the General Meeting on 20-22 May 2026, and the Executive Board has asked every member to register and vote. Most members will skim the headline page, note that the LIR fee held at €1,800, and move on. I see that miss every year.

I run an IPv4 marketplace. I read these reports the way an operator reads a renewal notice, looking for what I have to act on, where a journalist would read for a press-release angle. The interesting content sits below the highlight reel, in the allocation tables, the transfer infographics, and the line items behind the flat fee. Those are the numbers that decide what your address space is worth and how you should budget for it. This piece is a reading guide: which figures matter, where the report can mislead an unprepared LIR, and what the fee debate actually changes for a small registry.

Two facts frame everything below. The reports are online only, though you can download them to print. And the Annual Report deliberately sits apart from the Financial Report so technical activity and money are auditable separately. Keep them separate when you read them.

The flat fee is the headline, and it is not boring

The 2026 LIR membership fee is €1,800, unchanged from the previous year under Charging Scheme 2026 (ripe-848). Independent number-resource assignments carry a separate €75 charge, and an ASN assignment is €50. The structure is flat: every LIR pays the same base fee regardless of how much address space it holds. That single design choice is the most consequential thing in the Financial Report, and it cuts in a direction most coverage gets backwards.

A flat fee is regressive by size. A registry holding a single /22 pays exactly what a transit provider holding millions of addresses pays. If you are small, you are subsidising the large holders' share of the registry's running cost. That is the tension the 2027 charging debate is really about. Under the proposed Option A, presented by Simon-Jan Haytink of the Charging Scheme Task Force, the base LIR fee would rise to €1,894. It is a modest inflation-style bump. But it lands on every member equally, and a small LIR absorbs the same percentage increase as a provider with a thousand times the holdings.

My position: validate the €1,800 figure against the published Financial Report before you renew, and read the Option A proposal before the meeting rather than after the vote. The per-resource fees are stable and easy to forecast. The base fee is the only moving part, and it is the one you can influence with a ballot. The General Meeting decides this, and historically members do not rubber-stamp board recommendations on charging.

Read the allocation tables before you trust an adoption slogan

The Annual Report's resource statistics are where regional reality lives. As of 2026, North America holds the largest share of global IPv4 allocations at about 39.5%, with Asia near 26.9% and Europe near 23.9%. That distribution is why the transfer market flows the way it does and why a European LIR cannot assume its regional pool is the global picture.

On the protocol side, distrust any single adoption number. Global IPv6 adoption sits somewhere around 45-50%, yet IPv4 still carries roughly 55-70% of traffic. Those two figures describe the same network and look contradictory only if you forget that capability and usage are different things. A node can be dual-stack and still send most of its bytes over IPv4.

The spread underneath the average is wider still. Hyperscalers like Google report around 82% IPv6 adoption and China near 72%, while Africa lags at roughly 11% and around 63% of websites globally still serve no IPv6 at all. The growth rate tells the rest of the story. IPv6 adoption that climbed near 8% a year in the late 2010s has slowed to roughly 2-3% annually since 2023.

Read plainly, this is not a migration in progress; it is a long-lived dual-stack equilibrium. For an operator that means IPv4 is a line you keep paying for years yet, well past the point where you expected to be winding it down. Among Fortune 500 companies, about 76% still run primarily on IPv4. Public-facing services such as websites, APIs, and payment endpoints keep needing routable v4. That is the demand my business sees daily, and the report's own numbers explain why it persists.

What the report does not say about the leasing market

Here is where a careful reader has to stop relying on the document. The Annual Report covers allocations, assignments, and transfers, which is registry activity. It does not set or report secondary-market lease prices. External market trackers put IPv4 leasing around $0.38 to $0.50 per address per month in 2026, with demand concentrated in small, flexible blocks (/24 to /22) suited to geotargeting and short-lived deployments. Treat that band as market color from outside the report, never a RIPE figure, and never quote a tighter number than the range supports.

The leasing-versus-buying decision is the one the report can inform but not make for you. Leasing keeps the cost operational and the commitment short. Buying converts it to capital and locks in the asset, with the depreciation and resale risk that implies. Neither is correct in the abstract.

The mistake I watch operators make is treating leasing as always-cheaper. Over a multi-year horizon with static usage, cumulative lease payments can exceed a purchase price. Lease for flexibility and for clean blocks you need now; buy for the stable core you will hold for years. Run that arithmetic against your own time horizon before you commit either way.

Governance is the lever you forfeit by ignoring it

The reports exist to prepare members for the General Meeting, and the meeting is binding rather than ceremonial. At the May 2026 GM, 3,049 votes were cast. Voting requires registration in advance. Skip it and you are excluded from the quorum that ratifies the financial audit and the charging scheme. If the Option A fee increase concerns you, the report is your homework and the ballot is your only lever, and you have to claim it before the registration window closes.

Before the meeting, I work through the same short verification pass every cycle. It is faster as a checklist of what to confirm against what good looks like:

That last row matters more than it looks. A blocked repository is the false alarm that costs members their prep time every cycle, and it is almost always an egress rule rather than a missing document.

The reports also map K-root and AuthDNS nodes and surface new RIPE Atlas tooling. If you operate critical services, those maps are worth cross-referencing against your own reachability measurements, because the report tells you where the anchors are. It cannot tell you how your paths reach them.