Every project of any size has a date, and most of those dates are wishes. Q2 2027 is not a wish. It is the date by which Ireland Quantum has committed to stand up the country's first sovereign quantum compute facility, and it holds because we worked back from it rather than forward to it. The difference matters. Forward planning produces optimism. Backward planning produces a critical path, and a critical path is the only honest answer to the question of when something will be ready.

This essay walks through that path. What has to be true on a given week for the next thing to start. What has long lead times and therefore cannot be rushed by money. What could push the date to the right, and what we are doing now to keep each of those risks from becoming the one that does.

What "delivery" actually means

Before the timeline, the definition. Delivering a sovereign quantum facility in Q2 2027 means a physical site in Ireland, powered, cooled, network-connected, secured to the standard the workload demands, and running quantum compute hardware that Irish researchers, Irish public bodies and Irish-resident commercial users can access under Irish jurisdiction. Sovereign is the load-bearing word. The compute has to sit on Irish soil, under Irish law, with Irish operators in the room. That constraint shapes the entire critical path because it removes the easiest options — colocating in a hyperscaler region elsewhere in Europe, or renting time on someone else's machine and calling it ours.

Once you accept the sovereign constraint, the project decomposes into four parallel tracks that have to converge: site, power and cooling, hardware, and operating capability. Miss any one of them by a quarter and the whole thing slips. That is the discipline of a quantum facility timeline.

The critical path, in plain terms

The quantum project critical path looks roughly like this, working backward from a Q2 2027 go-live:

1. Acceptance and handover. The hardware has to be installed, cold, calibrated, and run through acceptance tests before any first user touches it. That window is measured in weeks, not days, and it cannot start until the room is ready.
2. Hardware delivery to site. Quantum systems do not arrive in a single crate. They arrive over a sequence of shipments — dilution refrigeration, control electronics, shielding, the processor itself — and each one has to land in the right order.
3. Mechanical, electrical and plumbing fit-out. The room has to be done before the hardware can be received. That means power distribution, cooling loops, vibration isolation, electromagnetic shielding, and the network drops, all signed off.
4. Building works and grid connection. The shell has to exist. The grid connection has to be live and stable. In Ireland in 2026 and 2027, the grid connection is not a paperwork item. It is its own project.
5. Site selection, planning, and procurement. Done. This is the work of 2025 and the first half of 2026, and it is the work that determines whether everything downstream is plausible.

Each of those steps has a finish-to-start dependency on the one above it. There is some overlap — you can be running hardware factory acceptance tests while the room is still being fitted out — but the merge point is unforgiving.

Where the long lead times actually live

The quantum supply chain is not the bottleneck most people assume. The processor itself is constrained, but the constraint is well understood and the order book is honest about it. The harder lead times are in the boring parts.

• Dilution refrigerators. The systems that take the qubits down to near absolute zero are built by a small number of vendors worldwide, in low volumes, with order books that price in the global build-out. You order these early or you wait.
• Specialist power and cooling kit. Switchgear, transformers, chillers at the right capacity. None of this is exotic, but all of it now sits behind the same European queue that data-centre operators are already in. Lead times that were weeks before 2022 are months now.
• High-purity helium-3. The working fluid for dilution refrigeration is not something you pick up locally. Supply is coordinated globally and you commit to it long before you need it.
• Grid capacity. The Irish grid is not short of plans. It is short of headroom in specific places at specific times. A connection agreement is not a connection.
• Cleared people. A sovereign facility needs operators who can hold the keys. That clearance pipeline is its own lead time, and it does not respond to money the way hardware does.

Notice what is not on that list: software. The control software, the user-facing scheduling layer, the billing and access control — those are real work, but they are not the items that decide whether the door opens in Q2 2027. They run alongside.

What could push the date to the right

I would rather write about risks honestly than pretend they do not exist. Here are the live ones, ranked by how much sleep they cost me, and what we are doing about each.

1. Grid connection slipping

This is the single largest risk. Ireland's grid is under load, and the timing of the energisation of a high-density site is not within the project's gift. Mitigation: early engagement, a confirmed connection path before any irreversible spend on the building shell, and a fallback design that can run on a smaller initial draw with a planned step-up. If we can run day-one workloads on less than the headline capacity while the larger feed comes online, the date holds.

2. Dilution refrigeration delivery slip

One vendor missing one shipment by one quarter and the calibration window collapses. Mitigation: orders placed early, deposits paid, and a relationship with the vendor that means we hear bad news in time to act on it rather than after the fact. We have also designed the room so that an alternative refrigerator footprint would fit without re-doing the slab.

3. Planning and statutory consents

Ireland's planning system is what it is. Mitigation: site selection done with planning risk as a first-order criterion, not a last-order one. We chose a route that does not depend on a marginal decision.

4. Specialist trades availability

The fit-out needs people who have done this kind of work before. There are not many of them in Ireland today. Mitigation: book the firms early, accept European mobility on key specialists, and pair them with Irish trades who are learning the work for the long run. The facility is not a one-off — the people we train on this build are the people who maintain it.

5. Hardware vendor roadmap shift

Quantum is moving. The processor we contract for in 2026 is not necessarily the processor that looks best in 2027. Mitigation: the room is built to be processor-agnostic within a reasonable envelope. We are buying capability, not a single chip. If a better generation lands inside the window, we can take it. If it does not, we ship with what we contracted for and upgrade later.

6. Funding rhythm

A long project with a fixed end date needs cash to land on the right weeks. Mitigation: staged commitments tied to gates that have already passed, not gates that are still ahead.

Why I still believe the date holds

Sovereign quantum delivery in Q2 2027 holds because the work that determines it has already been front-loaded. The hard procurement decisions are being made now, in 2026, when there is still slack in the schedule. The site logic is not contingent on a single planning decision. The hardware order is in the queue at the right point in the queue. The grid conversation is live, not aspirational.

It also holds because the team is small and the decisions are short. Most large infrastructure projects slip not because any one thing went wrong but because the path from problem to decision is too long. We have kept that path measured in days. When a vendor flags a slip, the decision about what to do about it happens that week. That is the only governance model I have ever seen work on a date this tight.

Finally — and this matters — Q2 2027 is a window, not a single Tuesday. Calling the date holds means delivering inside that window. Anyone who promises a specific day three years out is selling something.

What to do this week

If you are an Irish researcher, a public-sector technologist, or a commercial team whose roadmap quietly assumes you will be using quantum compute in 2027 and beyond, the work this week is not waiting. It is writing down, plainly, what your first three workloads would be, what data they touch, and what jurisdiction that data has to stay in. Bring that document to us. The reason we are building this on Irish soil under Irish law is so that you do not have to redesign your problem to fit someone else's region. At IMPT we are doing the same exercise in the open — naming the workloads we expect to put on the facility, what they need, and when. That list is what turns Q2 2027 from a delivery date into a useful one.