Public results. Named datasets. No percentage theater.

Every row below is something we will stand behind: a measured value, a public evidence anchor, and an honest scope. If it is not here, we have not published it as a claim.

~4 min read SuperCon · NASA PCoE Outcomes only

Public ledger

What we have measured and will defend.

  • Materials property accuracy

    R² 0.876

    Evidence

    Critical-temperature prediction on the public SuperCon benchmark

    Honest scope

    Competitive, not state-of-the-art. A triage tool, not a replacement for first-principles accuracy.

  • Explained variance vs baseline

    ~4×

    Evidence

    Same SuperCon task and inputs as a conventional statistical baseline (~R² 0.22 vs 0.876)

    Honest scope

    Fair conventional reference, not every possible competitor. Direction and magnitude, not a record claim.

  • Prediction latency

    ~0.02 s

    Evidence

    One material-property prediction on a single commodity board

    Honest scope

    Timing is for the prediction step; end-to-end throughput depends on your pipeline and hardware.

  • Screening throughput (derived)

    >1,000,000 / day

    Evidence

    Derived from ~0.02 s per prediction (~50 / s) on one commodity board

    Honest scope

    Throughput framing from the published timing — not a separate multi-day campaign measurement.

  • Battery early warning

    87–114 cycles

    Evidence

    End-of-life flagged ahead of a variance baseline on public NASA PCoE lithium-ion cells tested

    Honest scope

    Limited-cell result. Analytical signal for decisions; does not replace certified safety testing. Broader chemistry is a roadmap milestone.

  • Hardware class

    Commodity / edge

    Evidence

    Published timings and battery work run without a GPU farm

    Honest scope

    Describes the hardware class of these results — not a blanket guarantee for every future workload.

  • Facility instrumentation posture

    Passive / read-only

    Evidence

    Deployment principle for facility pilots

    Honest scope

    Trust and operations claim, not a performance percentage. Shadow / supervisory modes before control authority.

Boundary

What is not on this page — on purpose.

Directional charts (for example, public IEA-framed compute demand pressure) are context for why physics-first matters. They are not WaveCore performance claims. See Why now on the home page.

Frequently asked

Straight answers.

Why so few numbers?
Because we only publish what we can defend against a named public dataset or a clearly stated measurement condition. A short table of verified results is more useful than a long list of unverifiable percentages.
Can these results be reproduced?
The materials and battery figures are measured on public SuperCon and NASA PCoE data against defensible baselines. Qualified partners can go deeper under NDA; this page stays outcomes-only.
Where are the datacenter efficiency percentages?
Not here yet. Quantified datacenter efficiency figures are being verified against ground truth before we publish them as claims. We will not invent reductions to fill a table.
How do I see more than what is public?
Request technical diligence or investor materials under NDA via Contact. Methodology, fuller benchmarks, and unpublished partner work stay off the public site by design.

Where next

Technology See the evidence move Proof cards, charts, and how WaveCore deploys. Research Where we push — and stop Baselines, failure surfaces, inspectable ≠ exposed. Contact Go deeper under NDA Qualified partners get the fuller materials path.
Go deeper

Want the charts behind the table?

Technology walks the public evidence; Contact opens diligence under NDA.

See the technology Discuss a pilot