Anyone who has tried to build hard infrastructure in Ireland knows that the engineering is the easy part. The hard part is the cash. Quantum compute is no different, except that the equipment is colder, the power draw is steadier than people assume, and the funding stack has more moving pieces than a typical data centre because the technology still sits between research and commerce. Committing to deliver Ireland Quantum in Q2 2027 means being honest about that stack now — what is capex, what is opex, what comes from grants, what comes from equity, and what gets cut if a grant agreement slips by two quarters.
The shape of the spend
Quantum facility funding splits cleanly, at least on paper. There is the building and the fit-out — power, cooling, shielding, racks, security, the boring civil work that any high-spec compute site needs. There is the quantum hardware itself — the systems, the dilution refrigerators or photonic benches depending on modality, the control electronics, the calibration kit. There is classical compute alongside it, because no useful quantum workload runs without a meaningful classical co-processor and a network into it. And then there is everything that keeps the lights on: people, power, maintenance contracts, software licences, insurance, the slow drip of consumables.
The first three are capex. The last is opex. Founders get this wrong all the time, and it matters here because grants tend to fund the first category well, the second category sometimes, and the third category almost never. If you build a funding model that quietly assumes a grant will cover year-three salaries, you have built a model that will fail in year three. The rule I learned across two decades at Tesco, Dunnes Stores, and Oracle is unglamorous: model opex from cash you actually control, and treat grants as accelerants on capex, not as life support.
The IDA conversation
The IDA quantum conversation in Ireland is more mature than people outside the sector assume. The agency has watched the semiconductor and hyperscale data centre playbooks closely, and it understands that a sovereign quantum capability is not the same as another colocation hall in west Dublin. The questions they ask are the right ones. Where does the talent come from. What does the power profile look like. What is the export-control posture. Who are the anchor users. What happens to the asset if the operator falters.
What the IDA cannot do is replace a commercial customer base. Their support is real, and for a project of this kind it is meaningful, but it is structured around employment, capital investment, and regional development outcomes. So the conversation has to be framed honestly. We are not asking the IDA to underwrite a research lab. We are presenting a facility that intends to serve Irish and European users, that will employ engineers and operators in Tipperary, and that will draw private capital alongside public support. The grant ask is sized to what the agency's instruments can actually carry, not to what would be convenient for the cap table.
EuroHPC and Horizon Europe — the slower money
EuroHPC Ireland is the obvious anchor at European level. The Joint Undertaking has been buying and siting quantum systems across member states, paired with classical HPC, and Ireland's absence from that map is increasingly visible. A hosting arrangement, or a partnership with an existing EuroHPC node, is a cleaner path than trying to construct a parallel structure. The co-funding ratios are public, the procurement frameworks are published, and the timelines, while long, are predictable. Predictable matters more than fast when you are sequencing capex against a delivery date.
Horizon Europe quantum calls are a different animal. They fund work, not buildings. They are excellent for software stacks, error-correction research, hybrid algorithms, application development with industrial partners — the layer above the hardware. For a facility operator, Horizon money is best treated as a way to fund the user community around the machine rather than the machine itself. If you build a model that needs Horizon to pay for steel, you will wait a long time for steel.
The honest sequencing looks something like this:
- Equity and IDA-aligned support carry the building, the fit-out, and the first hardware tranche.
- EuroHPC participation expands the hardware footprint and brings the facility into the European compute fabric.
- Horizon Europe and national research funding, applied for jointly with universities and industrial partners, fund the workloads that justify the hardware.
- Commercial revenue from anchor users — including IMPT's own AI and travel workloads — covers a meaningful share of opex from year one.
The equity-and-grant blend
People sometimes talk about grants as if they were free money. They are not. Every grant euro carries reporting obligations, audit exposure, milestone constraints, and, in the case of EU instruments, a co-funding requirement that has to come from somewhere. If you stack three grants on a single project and each requires forty per cent match, you have not raised forty per cent. You have committed to raising the rest, on time, in cash, with the right legal structure, before any of the grant money lands.
The blend that works for a facility like this is equity-led, grant-accelerated. Equity sets the pace, defines the governance, and absorbs the risk that a grant decision slips. Grants compress the timeline, lower the cost of capital on the capex line, and signal to commercial customers that the project has institutional backing. Reverse the order and you end up with a project that is structurally fragile, because every delay in Brussels or Dublin becomes a delay in your own delivery.
For Irish tech grants in 2026, the practical advice I would give any founder reading this is to build a model that delivers a usable product on equity alone, then treat every grant as a switch that either pulls a milestone forward or expands scope. If the grant slips, the milestone slips, but the company does not.
What gets cut if grants take longer than they should
This is the question I spend the most time on, because it is the question that actually determines whether a facility opens on schedule. Grant agreements take longer than the public timelines suggest. State aid clearance takes longer. Co-funding partners need their own boards to approve. Procurement frameworks have appeal windows. None of this is anyone's fault — it is the cost of public money being public — but it has to be planned for.
The cut list, in order, looks like this:
- Scope of the second hardware tranche. One modality, one vendor, one system at opening, with the second system deferred to a later phase. This is uncomfortable but survivable.
- The research-grade visitor programme. Lovely to have, not core to revenue. Easy to reinstate when the grant lands.
- In-house software development beyond the integration layer. Buy more, build less, until the cash position recovers.
- The marketing and outreach budget. The facility's reputation will be built on uptime and published benchmarks, not on conference stands.
What does not get cut, ever, is power redundancy, security, the core operations team, and the commitments to anchor users. Those are the things that make the facility a facility rather than a press release. Cut them and you have nothing worth funding in the first place.
Tipperary as a feature, not a footnote
Siting in Clonmel is a deliberate choice and it changes the funding conversation in useful ways. Regional development weighting in IDA instruments is real. Power availability outside the Dublin grid pinch is real. The talent question is harder, and I will not pretend otherwise — we will have to bring people in, and we will have to grow people locally, and both take time. But the case for a sovereign Irish quantum capability is stronger, not weaker, when it is not stacked on top of the same infrastructure that every hyperscaler is already competing for. The Suir valley has water, power, and a workforce that has built complex things before. That is the starting position.
What we are doing about it this week
At IMPT we are treating Ireland Quantum as a delivery commitment, not an aspiration, which means the funding work is happening in parallel with the engineering work rather than after it. This week that looks like finalising the capex model against three grant scenarios — full stack, partial stack, equity-only — so that the Q2 2027 date holds in all three. If you are a founder looking at a similar build, or a policy reader trying to understand why these projects take the shape they do, the practical step is the same: write the equity-only version of your plan first, in detail, and only then layer the grants on top. If the equity-only version does not stand up, no grant stack will save it. If it does, the grants make a good project faster, and that is the right order of operations.