Energy-efficient cryptography, performance per watt
In a fixed power envelope the useful measure is cryptographic operations per watt, not operations per second. Post-quantum algorithms cost more energy per operation than the ones they replace, so the same workload draws more power after migration than before it.
Why the envelope, not the ceiling
Data centre capacity is increasingly limited by power and cooling rather than by space or by available silicon. A rack has a budget, and every watt spent on cryptography is unavailable for the work the rack exists to do.
At the edge the constraint is harder still: a cabinet or a device may have a few tens of watts total, and cryptography competes with the function.
The direction of travel
Cryptographic energy per transaction fell for years as CPU instructions absorbed the work. The post-quantum transition reverses that for the first time in a long while: operations cost more, hybrid modes do two of them, and key material is larger to move.
The consequence is straightforward and unwelcome — the same session volume draws more power after the migration than before it, on hardware chosen before anyone measured.
Why dedicated hardware helps here specifically
A general-purpose CPU pays for generality on every operation: instruction fetch, decode, scheduling and cache traffic that a dedicated datapath does not have. For a regular, repeated operation like an NTT the difference in energy per result is substantial.
That is the same reason fixed-function hardware wins on efficiency generally, and it applies as much to entropy generation as to arithmetic — a source that produces entropy directly costs less energy per bit than a CPU stirring a pool.
What to measure
- Joules per operation, by operation type, not average power.
- Idle draw as well as active, since an accelerator that is idle most of the time still costs its static power.
- Whole-system energy for the workload, not the component in isolation — offload that halves CPU energy and adds more elsewhere is not a saving.
- Energy per operation before and after the post-quantum change, on the same hardware, which is the number that makes the case internally.
Why it is worth arguing now
Because the trend only moves one way. Session volumes rise, per-operation cost has risen, power budgets are fixed, and the gap between those has to be closed by efficiency somewhere.
Keep reading
- Infrastructure and efficiency
Edge and telecom cryptographic acceleration
At the edge, cryptographic load is distributed across many small sites rather than concentrated in a few large ones, and latency and autonomy matter more than aggregate throughput. That changes the answer: many modest local capabilities rather than one large central one.
- Infrastructure and efficiency
TLS, IPsec and PKI acceleration
Cryptographic load in infrastructure accumulates at TLS termination, IPsec tunnel endpoints and PKI signing. Post-quantum handshakes increase the per-session cost at the same session volume, so a tier sized before the migration may not hold after it.
- Infrastructure and efficiency
Innovative Strategies for the New Era of High Frequency Trading
Leveraging these cutting edge High Frequency Trading Strategies to strategically navigate the complexities of the financial markets. Read on to learn more!