Investor Brief · The Founders' Group · Ben Wiggins

Catalyst‑Q

A utility-scale quantum execution substrate — available for verification today.

Aligned to DARPA's Quantum Benchmarking Initiative (QBI) Stage A · DARPA-PA-26-02-02 · abstracts on a rolling basis through July 31, 2026.

virtual · distributedtopological · bounded-widthpublic benchmark record
Strategic Innovations · Travis Crew · 2026
Why Now — The Window Is Open

DARPA is actively funding a path to a utility-scale quantum computer.

"If an approach to quantum computing is discovered that enables utility-scale quantum computing much faster than anticipated, DARPA is interested in the immediate and timely verification and validation of the approach's viability — performed in parallel with ongoing R&D efforts." — DARPA-PA-26-02-02, QBI Stage A, §B Introduction

That clause was written for an approach exactly like ours. We are the faster-than-anticipated approach — and we are ready for V&V now.

Rolling abstracts through Jul 31 2026 · full proposals Sep 30 2026 · new entrants explicitly encouraged. Solicitation on SAM.gov ↗

What DARPA Is Buying

QBI in one glance

  • Goal: a utility-scale, fault-tolerant quantum computer (USQC) and workflows with quantum compute steps.
  • USQC, defined: "one whose computational value exceeds its costs."
  • Two abstract criteria: Technical Plausibility, then Technical Ability — on the proposed timeline.
  • Award vehicle: Research OT (10 U.S.C. §4021) or Prototype OT (§4022).

Stage A · key dates

Abstracts (rolling)Jul 31 2026
Full proposalsSep 30 2026
Stage A term / ceiling6 mo · $1M
Technical POCMTO · QBI
The Problem — A Double Deadlock

Everyone else is waiting a decade for the hardware.

  • The physical-qubit wall. Superconducting, trapped-ion, neutral-atom stacks are throttled by noise and coherence; fault tolerance is widely predicted to be decades away.
  • The simulation wall. Classical statevector emulation scales as 2n — tracking entangled states past ~50 qubits needs exabytes. Utility-scale simulation is "physically impossible" on conventional infrastructure.

So the market rewards whoever can execute useful quantum workloads today — not another roadmap promise.

The Solution

We didn't wait for the hardware. We changed the representation.

Catalyst-Q executes gate-model and optimization workloads on a bounded-width phase representation — never materializing the 2n statevector, never depending on a fragile physical qubit stack.

Virtual — runs in software on the cloudDistributed — stateless, deploy anywhereTopological — lossless entanglement bindingBounded-width — polynomial, not 2ⁿ

Honest framing: this is a virtual quantum-computation substrate on commodity classical infrastructure, distributed through Cloudflare's edge — not a cryostat. That is precisely what makes it available for immediate V&V, and what maps it onto a room-temperature photonic hardware path in Stage C.

How the wall dissolves

Memory-wall eliminated. Only time scales — and only for genuine volume-law.

  • Zero materialization. Exact amplitudes, marginals & observables read by targeted resonance — the joint 2n state is never written down.
  • GHZ chain, 10,000 qubits → branch width 2, built in 1.64 s; flat in entanglement depth.
  • 5,000 T-gates (non-Clifford) → width 2 — magic states are free phase.
  • Exact vs. reference: GHZ / QFT / Grover match Qiskit statevector bit-for-bit where Qiskit caps out near 24–50 qubits.

The honest boundary (we say this out loud)

Exactness and bounded memory hold for every circuit. For all-to-all volume-law circuits, time scales — the same #P-hardness every exact method faces. We eliminate the memory wall that stops statevector at ~50 qubits; we do not claim to repeal complexity theory. No supremacy claims. No "simulate any circuit for free."

Utility Scale — As DARPA Defines It — We Already Fit

Value exceeds cost, today, on a public record.

1.0
Mirror Circuits exactness (to width 64)
1.0
QML Kernel accuracy (10–50 qubits)
0.7735
Linear-Ramp QAOA approx. ratio
5.97e‑11
EPLG leakage, 100-qubit chain

These are not slideware. They are third-party, open benchmarks on the Unitary Foundation's Metriq record — the industry's neutral scoreboard.

Press ↓ for the three proof drill-downs

Proof 1 · DARPA's own yardstick

"Computational value exceeds its costs."

A utility-scale run is one where the answer is worth more than the compute to get it. Catalyst-Q returns exact results for workload classes — Mirror Circuits, QML kernels, QAOA — at widths and depths where physical qubits are still noisy and classical statevector has already run out of memory. Value delivered, cost bounded. That is the definition, met.

Proof 2 · The public record

Metriq PR #461 — 13 scoring buckets, device catalyst_q_v3_1

Submitted to the Unitary Foundation (June 18, 2026)

Mirror Circuits (to width 64, 4 layers, 1000 shots)1.0 exact
QML Kernel (scaling qubit sizes)1.0 accuracy
Linear Ramp QAOA (10-qubit graph)0.7735 ratio
EPLG (100-qubit chain)5.97e-11

Public, third-party, reproducible — the same repository DARPA and the field already use for cross-platform evaluation. Anyone can check it.

Proof 3 · Every result carries its own receipt

Signed, replayable .rain certificates.

Each execution emits a hash-chained, re-runnable evidence packet. A verifier re-executes the certificate and gets the identical answer — deterministically. This is not "trust our demo." It is verification built into the substrate — which is exactly the muscle DARPA's IV&V team needs.

IP boundary preserved: certificates prove what was computed and let it be re-checked, without disclosing the routing logic or phase-binding internals. Maps cleanly to QBI Category 2 (V&V) data rights.

Why Us — DARPA Wants Two Things. We Are the Only One Who Brings Both.

A utility-scale substrate and the instrument to verify it.

1 · The thing to be verified

A running utility-scale quantum-computation substrate — live, distributed, benchmarked — available for immediate V&V today. That is the "faster-than-anticipated approach" §B was written for.

2 · The instrument that verifies

The same targeted-read engine is a verification instrument for other performers' circuits — an exact, memory-bounded oracle plus signed replay. DARPA's Stage-C IV&V mission, in a box.

To our knowledge, no other approach is simultaneously (a) a utility-scale execution substrate ready for V&V today, and (b) a verification instrument for the rest of the field. That dual fit is the moat.

The Route — Stage A → B → C, Accelerated

Software-validated first means we start already de-risked.

Stage A · 6 mo · ≤ $1M
Plausibility
Concept design report + IV&V review. Our engine already exists in executable form — we validate what's live.
Stage B · 12 mo · ≤ $5M
R&D Plan
Map the validated virtual substrate onto a room-temperature photonic realization path; reduce build risk.
Stage C · ~36 mo · ≤ $300M
Validation & Co-Design
Embed the already-perfected logic model into specialized photonic silicon; hardware verification against benchmark targets.

Most performers spend Stage A trying to show a concept could work. We spend it verifying a substrate that already does — and Stage A milestones may be completed early to open Stage B ahead of schedule.

Press ↓ for the milestone & payment schedule

Stage A milestones & payments

How the first $1M is structured (DARPA Table 2)

MilestoneDeliverableDueDARPA funds
A.1.a · KickoffMilestone report + government technical reviewMonth 1up to $150,000
A.1.b · Concept Design ReviewIn-person CDR on the draft concept reportMonth 4up to $500,000
A.1.c · Final Concept Design ReportRevised USQC concept per IV&V feedbackMonth 6up to $350,000

Our acceleration lever: because the execution engine is already validated in software, the Month-1 kickoff can ship an evidence-replay harness the IV&V team runs themselves — front-loading confidence and opening the door to early Stage B negotiation.

Traction — Shipped, Downloaded, Public

No revenue yet. But the substrate is already in the world.

3,170+
SDK installs / month (PyPI, both packages)
2,533
catalyst-brain installs, last 30 days
637
catalyst-q installs, last 30 days
13
public Metriq scoring buckets (PR #461)
  • Live SDKs: pip install catalyst-q · pip install catalyst-brain — shipping, versioned, documented.
  • Public benchmark footprint on the Unitary Foundation's Metriq — third-party, not self-reported.
  • Working products on the substrate already deployed on Cloudflare (agentic memory, solvers, verification).

Honest status: pre-revenue. The path to revenue is a DARPA Stage A OT ($1M non-dilutive) plus commercial substrate licensing — download momentum and a public benchmark record are the leading indicators. catalyst-q on PyPI ↗

The Ask

Back the abstract that puts a working utility-scale substrate in front of DARPA.

Rolling deadline — abstract due July 31, 2026

  • Near-term: a bridge to finalize the Stage A abstract (Attachments A & B), SAM.gov registration, and the IV&V evidence-replay harness.
  • The prize: a Stage A OT — up to $1M non-dilutive — with a de-risked line to Stage B ($5M) and Stage C (up to $300M).
  • Your role, Ben: narrative sharpening, warm intros, and the credibility of The Founders' Group behind a first-of-its-kind submission.

Why this is a rare shape of bet

Non-dilutive government capital, a public benchmark moat, a live product with real installs, and a clause in the solicitation that reads like it was written for us. Small check now; asymmetric line to $300M and a category-defining position.

Let's Verify the Future — Together

The only approach ready for DARPA's V&V today.

Travis Crew · Strategic Innovations

sales@strategic-innovations.ai

pip install catalyst-q · Metriq PR #461 · DARPA-PA-26-02-02

All figures from public sources: the DARPA QBI solicitation, the Unitary Foundation Metriq record, and live PyPI statistics. Virtual-substrate framing is deliberate and defensible under technical review.

← → columns · ↑ ↓ detail · F full · P print
Catalyst-Q · QBI