Cutaway illustration of the Blitz M.2 accelerator card: aluminum enclosure cut open to expose the FPGA, with an expanded die view showing the K1 verify engine, the RISC-V runtime auditor running libsecp256k1 in software, and the PCIe DMA block, plus flash/boot (golden and A/B images), power and thermal monitoring with throttling, and JTAG support blocks
Fig. 1 Blitz, the M.2 accelerator card, cut away — the K1 engine and its runtime auditor on one FPGA
297,100
Schnorr verifications / s
host-measured · Ti375 FPGA · 4 K1 cores
91.3%
RTL line coverage
1,410/1,544 lines, from our repos
20/20
formal checks green
SymbiYosys, incl. anti-vacuity

Generated from our repositories, commit-stamped — not marketing copy. See the full panel →

Why this exists

The bottleneck nobody solved

Every transaction a blockchain accepts rests on signature verification — the costliest step of validating a block. For fifteen years the ecosystem has worked around that cost rather than removing it. Removing it is what Node Labs does.

THE PROBLEM

Nodes already skip it

Checking every signature from genesis is expensive enough that Bitcoin Core skips historical blocks by default (assumevalid) — the ecosystem's most conservative software already treats full verification as too slow to insist on. With cheap verification, you check everything.

Why not just use a CPU? →

THE CEILING

It's why blocks stay small

Purpose-built silicon made mining ∼600,000× more efficient in fifteen years. Validation never got that treatment — it stayed on general-purpose CPUs, and Bitcoin's capacity stayed where it was: roughly seven transactions per second. A chain can only scale as far as ordinary machines can afford to verify.

The mining-hardware precedent →

THE ANSWER

Point silicon at verification

The same trajectory, aimed at the operation networks actually depend on: one M.2 card verifies 297,100 signatures per second — measured on real hardware, running live below. Enough that verification is never the reason to skip checks or keep blocks small.

The Blitz card →

Live from the bench

A real card is running right now

The meter beside this text is not a mockup. It streams 20×/s from a real Ti375 FPGA on our bench in Vancouver, verifying Schnorr signatures as you read this — including checks mirrored from a live node on Nexa, the big-block chain we partner with to prove our silicon in production. If the bench is offline, it says so.

Most of the time the needle sits near idle — today's networks barely register against hardware like this. That headroom is the point: capacity has to exist before the traffic that needs it can, and validation should never again be the reason a chain stays small.

Push load onto it yourself →

live — streamed from the bench in Vancouver. Open the full viewer

What we build

One engine. Two ways to get it.

Our business is licensable silicon IP for cryptographic acceleration. Our hardware exists to prove that IP on real workloads — and to put it in your machine.

Path to silicon

FPGA-proven today. ASIC next.

NOW

Field validation on FPGA

K1 verification cores running on Blitz hardware, exercised on real cryptographic workloads — including live testing on the Nexa network, where our founder is one of the network's designers.

NEXT

130 nm trial tape-out

A trial core through the open-source OpenROAD flow to de-risk the path to production silicon.

THEN

Advanced-node production run

The secp256k1 point multiplier core — working in hardware today — fabricated at a leading foundry.