Ageing sits upstream of dementia, arthritis and much of oncology. Treating it as a risk factor rather than background noise is the only lever that acts on several diagnoses at once.
Four tracksRung 01 280.000 DKKFull build 2.4M DKKZero EgressOn-premise
Track 03 · Capital
Epivora Labs
Horsens · DK
Fueling
The Longevity
Dividend
Capital allocation at Epivora Labs is ruthlessly efficient. We own the compute, we don't pay cloud tax, and we automate execution. We are raising capital to fund physical rungs of capability and sustained model inference.
Four rungs · each stands alone · quarterly capability reporting
01The Rungs
Capital converts to capability,
not to rented time.
Each rung is a discrete piece of physical capability with a defined boundary. Fund only the first and the lab still gains something permanent it did not have before. Expand a rung to see exactly what the money buys and, just as importantly, what it does not.
- Unlocks: in-house long-read sequencing, and with it direct methylation calling from native DNA.
- Every hypothesis the models generate becomes testable on site instead of queued with an external provider.
- Turnaround on a validation run drops from weeks to days, which changes how many hypotheses are worth testing at all.
- Includes the instrument, initial flow cells, extraction bench and installation.
Does not cover
Automated preparation. Library prep stays manual at this rung, so throughput is capped by available hands — but the loop starts producing its own data.
- Unlocks: automated sample and library preparation with a HEPA enclosure.
- Reproducibility stops depending on who was on shift. Protocols become versioned parameters rather than lab-book notes.
- Overnight and weekend runs become possible, roughly tripling usable instrument hours without adding staff.
- Assumes Rung 01 is in place; together they make stages 02 and 03 of the loop physical.
Does not cover
Full hand-off between instruments. A person still moves plates between the handler and the sequencer until Rung 03 links them.
- Unlocks: the closed loop. Sequencing, robotics and the model layer linked into one chain with no manual handover.
- Scheduling, error handling and instrument telemetry wired into the existing pipeline so failed runs are caught by software.
- Includes twelve months of consumables, reagents and inference budget — the loop has to run long enough to validate itself.
- This is the rung where the lab stops being a collection of instruments and becomes a system.
Does not cover
Redundancy. A single failure on a critical instrument still stops the loop until it is repaired.
- Unlocks: all four research tracks running in parallel rather than in sequence.
- Duplicated critical hardware, so a single instrument failure degrades throughput instead of halting it.
- Expanded GPU capacity for local inference and fine-tuning, reducing dependence on external model APIs.
- Extended runway so the programme is judged on results rather than on the fundraising calendar.
Does not cover
Clinical trials. This funds discovery and preclinical validation. Anything beyond that is a different conversation with different partners.
02Operational runway
Where the money goes
every single month.
Owning the cluster removes the largest recurring cost in AI-heavy research, but it does not remove all of them. This is the honest shape of the burn: mostly consumables and inference, with no line item for office design.
AI token usageExternal API calls for frontier models we cannot run locally. Falls as GPU capacity grows.
34%
Reagents and consumablesFlow cells, kits, plates, tips. Scales directly with how many experiments actually run.
26%
Power and coolingThe cluster runs continuously. Danish electricity is the price of keeping data on site.
18%
Secure facility overheadRent, access control, physical security, network. The perimeter that makes Zero Egress real.
13%
Insurance and complianceInstrument cover, liability, data protection and the documentation partners require.
9%
Allocation is indicative and shifts with activity. Actual monthly figures, supplier quotes and historic burn are in the funding pack.
03Return on capital
The people who will fill
the care system in 2045
are already born.
The number is fixed. The only variable left is how many of their years are healthy ones. Health systems are built to repair damage after it happens, and that model does not scale once the curve turns. The alternative is not more hospitals — it is moving the intervention back to the cell, before the pathology is established.
This is the Longevity Dividend: every healthy year recovered is both a person kept out of the system and a working life society gets to keep. Compressing morbidity is the rare intervention that improves individual outcomes and public finances with the same mechanism.
No cloud tax, no rented compute. Funding buys hardware that keeps working after the grant period ends — a different asset profile from a budget spent on API calls.
On-premiseFour tracks share one pipeline, one index and one pool of compute. Moving a finding from one track to another costs nothing — the structural advantage a small lab actually has.
Shared pipelineQuarterly capability statements: what was funded, what was built, what it produced, what failed. Including the failures — a research programme that only reports wins is not reporting.
Quarterly—Three ways to build this lab
The Brain
Enterprise compute, memory, storage and networking for on-premise training and inference.
Open this track → The HandsAutomated liquid handling, sequencing and assay hardware to close the research loop.
Open this track →The Fuel
Capital for physical rungs of capability and sustained model inference.
You are here