Summary

  • RIPE NCC appointed Sjoerd Wolthers permanent CTO from 1 September 2026 after a recruitment process involving around 270 applicants; he has served as interim CTO for about a year.
  • Draft minutes from the latest publicly listed Executive Board meeting record a reasonable infrastructure estimate of an additional €5m CAPEX and €7.2m OPEX over five years, including four additional FTEs.
  • The same June minutes say requirements and solutions still needed refinement and that the Board required more information before taking a firm resolution. The estimate is not an approved budget or final cost.
  • RIPE NCC should publish a versioned capital-decision receipt that separates the appointment from project approval and joins any eventual spending decision to service baselines, cost ranges, funding, milestones, acceptance tests and rollback conditions.

A job announcement closes one gate

On 11 August, RIPE NCC announced that Sjoerd Wolthers would become its permanent chief technology officer on 1 September. The organisation said it had considered around 270 applicants through several selection rounds. A fuller announcement to the RIPE List describes multiple interviews, a leadership and capability assessment, and a written assignment. Wolthers had already spent about a year as interim CTO after earlier technology leadership roles at KPN, XS4ALL and FD Mediagroup.

That is a consequential appointment. RIPE NCC's technology division supports systems that operators across its service region use to register number resources, maintain authoritative data and secure route origins. The appointment also arrives while the organisation is considering a multi-year rebuild of the foundations beneath several of those services.

It is tempting to combine the two stories: a permanent technology chief has been selected, therefore the infrastructure programme has its mandate. The public record does not support that shortcut. Hiring a person and authorizing a capital programme are different decisions, taken through different controls and producing different evidence.

The appointment says who will lead the technology function. It does not say that the Executive Board approved a particular architecture, released money from the reserve, accepted four new posts, selected suppliers or authorized migration. Keeping that distinction clear is not a procedural technicality. It protects the incoming CTO as much as it protects members: delivery can be judged against a decision actually made, rather than a mandate inferred from a job title.

The project record is still conditional

RIPE NCC began exposing the infrastructure question before the appointment. Final minutes of the 191st Executive Board meeting, held on 26–27 March, say the current platforms were too old for major restructuring. Management's preferred path was to build a new platform in parallel and migrate services rather than keep reworking old foundations. The same record describes the budget view as significant and rough and notes that a full design did not yet exist.

At RIPE 92 on 20 May, Managing Director Hans Petter Holen made the scope more concrete. He named the RIPE Database, RPKI and the LIR Portal and registry as core services that should move onto a new foundation. He described a programme running from 2026 to 2028, using external expertise to accelerate design and construction while retaining the ability to operate the result internally.

The presentation was also more nuanced than a simple “leave the cloud” slogan. Holen spoke about reducing dependencies that could threaten continuity across geopolitical or supplier failures, but said RIPE NCC would not eliminate cloud services or external providers. The aim was control and independent operation for core capabilities, not technological isolation.

Draft minutes of the 194th Executive Board meeting on 22–23 June carry the decision trail forward. They say architectural choices had been made and others remained outstanding. Requirements still covered data-centre configuration and location, application workstreams, security, storage and backup. Board members asked what capacity was needed and what the proposal included so that they could understand the project's size.

The minutes then record the number that will attract most attention: a reasonable estimate of an additional €5m in capital expenditure and €7.2m in operating expenditure over five years. The operating estimate included four additional full-time-equivalent posts. The minutes immediately qualify the figures. Requirements had to be finalised, solutions approved and estimates refined. Negotiations were continuing in some areas.

Funding was also unsettled. Options included use of the reserve, increased member income and cost reductions. Leasing was to be investigated; a loan was not being considered. Management still needed to identify which existing costs the project would replace and improve capacity and asset management for recurring equipment renewal.

The Board's recorded conclusion is therefore as important as the estimate. It said the requested information was needed before any firm resolution could be made. The aim was to provide that detail and convene an additional meeting in the following weeks.

€12.2m is arithmetic, not authorization

Adding the two June estimates produces €12.2m over five years. That arithmetic is useful only if the label travels with it. It is not an approved budget, a procurement award, a final project cost or a forecast that RIPE NCC will necessarily spend every euro. It combines different accounting categories across a five-year horizon and includes staffing that may change when scope and design change.

The current public 2026 Board-minutes index, checked on 29 August, lists meetings only through the 194th meeting in June. No later minute or resolution appears there. That observation defines the edge of the public evidence checked; it cannot establish that no internal meeting happened or that no later decision exists elsewhere.

This is precisely why a durable decision receipt matters. A news announcement can become current before the underlying governance trail catches up. Readers should not be required to infer whether a later meeting occurred, whether an estimate changed or whether a delegated approval was used by comparing a job announcement with an old draft minute.

The right correction is not to delay the appointment or force infrastructure design into public debate. It is to publish a compact, versioned record whenever the capital gate moves.

What a capital-decision receipt should contain

The first field is authority. The record should name the decision, its version and date, the body that took it, and any delegated limit used. If the Board authorizes only design work or a first tranche, the receipt should say so. If it approves a range subject to later gates, those gates should be visible.

The second field is scope. “Infrastructure” is too broad for accountability. The receipt should identify the services and capabilities included and excluded: for example, the RIPE Database, RPKI, LIR Portal and registry functions, shared storage or identity dependencies. Security-sensitive topology, provider bids and physical locations can remain confidential.

The third field is a before-state. A migration cannot demonstrate improvement if the starting condition disappears. RIPE NCC should preserve service availability, recovery and restore-test results, capacity, deployment lead time, incident exposure and comparable operating cost before material change begins. A baseline is not a promise of perfect service; it is the denominator for later claims.

The fourth field is the estimate with its grammar intact. Capital, operating, staffing and external assistance should be separated. Each range should state what it includes, what it excludes, its pricing date and the demand assumptions behind it. Existing expenditure that will be retired must be distinguished from costs that continue and from temporary duplication during parallel running.

The fifth field is funding. Members should be able to see whether a tranche comes from the reserve, future charges, reprioritised activity or savings, and what happens if reserve coverage crosses a stated threshold. That does not give members an operational veto. It gives them the information needed to distinguish resilience investment from an unbounded call on future income.

Finally, the receipt needs delivery evidence: milestone owners, acceptance tests for each core capability, reconciliation during dual running, rollback conditions, material variance thresholds and a reporting cadence. “Platform delivered” is not an acceptance test. A restored RIPE Database, a validated RPKI publication path and a completed registry transaction are observable capabilities.

What the appointment does not prove

The public record gives no basis for attributing RIPE NCC's inherited architecture or earlier cloud strategy to Wolthers. Serving as interim CTO for the past year made him part of the current design work; it did not make him the author of every prior decision. Nor does his permanent appointment transfer the Board's budget authority into the CTO role.

The evidence also does not select an architecture for RIPE NCC. Multiple independent installations, cloud services, leased infrastructure and internally operated platforms can be combined in many ways. The governance question is whether the chosen arrangement satisfies explicit resilience, security, maintainability and cost tests—not whether it fits a fashionable label.

Nor should the project be justified retroactively by attaching every service incident to old infrastructure. The March minutes, May presentation and June minutes articulate a prospective case. Individual incidents require their own causal records. A capital receipt should not become a container for blame.

The incoming CTO therefore inherits two valuable things: a clearer public statement of the infrastructure problem, and an unresolved decision that can still be documented well. RIPE NCC has selected the person. Its next public record should show what, exactly, that person has been authorized to deliver.

Sources