IOT News
Post-Quantum Security Starts Moving Toward IoT Hardware
Published by IOT San Jose Research & Editorial Team

A smart meter installed this year may still be reporting in 2040. An industrial sensor bolted to a pump can outlast three generations of laptops. Infineon, announcing a post-quantum security controller in October 2025, pointed out that IoT devices such as industrial sensors, smart meters, and control units often operate for 10 to 20 years. That lifespan is the heart of the post-quantum problem. Nobody knows exactly when a quantum computer will be able to break today's public-key cryptography, but a device shipped now has to stay secure through years when one might exist.
For IoT, this is not a distant research question. It is a hardware planning question, because the cryptography that protects a device lives in silicon, in secure elements, and in firmware that is hard to change once deployed. Chip makers have started shipping parts that carry the new algorithms, and regulators are beginning to ask product teams to show a plan.
What Is Actually at Risk
Quantum computers threaten one family of cryptography, the public-key methods that underpin key exchange and digital signatures, such as RSA and elliptic curve schemes. Symmetric encryption and hash functions, such as AES and SHA-2, are affected much less, and the usual response is to use longer keys. The practical exposure for IoT falls into two buckets.
- Signatures that prove authenticity. Secure boot, firmware updates, and device identity depend on signatures. If an attacker can forge a signature, they can push malicious firmware that the device accepts.
- Key exchange that protects secrecy. Encrypted traffic that is recorded today can be decrypted later if the key exchange is broken. This harvest-now, decrypt-later risk matters for data that must stay confidential for many years, such as medical or industrial design data.
Signature forgery is the sharper concern for devices, because a forged update can compromise a whole fleet. Confidentiality matters most for sensitive streams. Both need attention, but the priorities differ by product.
The Standards Are Settled Enough to Build On
In August 2024, the US National Institute of Standards and Technology published its first post-quantum standards. ML-KEM, in FIPS 203, handles key establishment. ML-DSA, in FIPS 204, and SLH-DSA, in FIPS 205, handle signatures. NIST also selected a further key-establishment algorithm, HQC, in 2025 as a backup. Stateful hash-based signature schemes, LMS and XMSS, were already standardized and are especially suited to signing firmware. Government timelines add pressure. US guidance for national security systems calls for quantum-resistant algorithms in new acquisitions on a schedule that begins in the next few years, and European authorities have published coordinated roadmaps that start the transition by the end of 2026 for the most critical systems. Exact dates differ by document and keep being refined, so product teams should check the current text rather than rely on a summary.
In Europe, the Cyber Resilience Act adds a second driver. Its vulnerability reporting duties start in September 2026, and the main obligations follow in December 2027. The Act does not name quantum-safe algorithms, but it requires secure-by-design products and a supported update path, both of which a post-quantum plan helps to satisfy.
What Chip Makers Have Shipped
| Approach | Example | What it provides | Best for |
|---|---|---|---|
| Software library on a standard microcontroller | STMicroelectronics X-CUBE-PQC for STM32 | LMS, XMSS, ML-KEM, and ML-DSA, with the chip's crypto accelerator reducing the cost | New designs where flash and RAM allow |
| Certified secure controller | Infineon SLC27 with a Common Criteria certified library including ML-KEM and ML-DSA | Side-channel and fault-attack resistance, in-field updates | Meters, industrial sensors, secure elements |
| Dedicated post-quantum secure element | SEALSQ QS7001 and similar parts | Hardware root of trust with quantum-resistant algorithms | Devices that need a separate, tamper-resistant trust anchor |
| Hybrid classical plus post-quantum | Combined key exchange or dual signatures | Security if either algorithm survives | Transition period for most products |
STMicroelectronics notes that its library works alongside its existing cryptographic package and is validated through NIST's algorithm validation program. It also states the trade-off plainly: post-quantum algorithms carry a performance penalty and higher memory requirements, and mixing them with classical cryptography can soften the impact. That candor is useful. A constrained device that already struggles for memory may not have room for the new algorithms without a redesign.
Why Hardware Is the Hard Part
Software teams can update a library. Hardware teams cannot add flash to a shipped board. Three hardware constraints decide how easy the move will be.
- Memory. Post-quantum keys and signatures are larger than elliptic curve equivalents, which affects flash, RAM, and the size of messages over narrow links such as LoRaWAN or NB-IoT.
- Compute and energy. Some algorithms run slower on small cores, which affects battery life and boot time. Accelerators help, which is why chip support matters.
- Immutable roots of trust. If the first-stage bootloader only verifies a classical signature and cannot be updated, the device's trust chain has a permanent weakness. Hash-based signatures such as LMS can be added at the root of trust today for this reason.
These constraints connect to broader hardware planning trends. Our coverage of chiplets and RISC-V in IoT hardware noted that dedicated security chiplets with hardware roots of trust are becoming a baseline requirement, and the discussion of smaller, faster edge AI chips shows how memory budgets are already under pressure.
A Worked Example: A Smart Meter Maker's Decision
Consider a South Bay company designing a utility meter for a ten-year contract that may be extended to twenty. This is an illustrative scenario. The firmware signing key currently uses an elliptic curve algorithm. The team asks three questions. First, can the root of trust be updated in the field? If not, the design should add a hash-based signature check now, since LMS public keys are small and the verification is lightweight. Second, does the secure element support ML-KEM and ML-DSA, or does it have a certified roadmap? Third, can the cellular link carry larger certificates without exceeding data plans?
The answers lead to a plan. Add LMS verification in the bootloader for the first release, choose a secure element with certified post-quantum libraries, and design the update mechanism so algorithms can be replaced later. That last property is called crypto agility, and Infineon lists in-field updates and crypto agility among its controller's features. It may matter more than any single algorithm, because standards and recommendations will keep evolving.
Harvest Now, Decrypt Later: Who Should Worry
The harvest-now, decrypt-later idea deserves a plain explanation. An adversary with enough storage can record encrypted traffic today and wait. If a capable quantum computer arrives later, the recorded traffic could be opened. The risk therefore depends on how long the data stays sensitive. A temperature reading from a warehouse loses its value in minutes. A patient record, a proprietary manufacturing recipe, or the design data for a new chip can stay valuable for decades.
A practical rule is to rank data streams by their useful secret lifetime and prioritize the longest. For devices that only send routine telemetry, authentication and signature integrity matter more than long-term secrecy, and the first step is protecting firmware signing. For devices that carry sensitive records, such as medical monitors or industrial equipment in a competitive process, adopting a quantum-resistant key exchange earlier is justified. A hybrid approach, combining a classical and a post-quantum exchange, lets teams start without betting everything on a new algorithm.
This risk framing also helps in conversations with executives. The budget request is not for a theoretical future threat. It is for protecting assets whose value outlasts the time before the threat may arrive.
Implementation Advice
- Inventory cryptography. List every place a device uses public-key cryptography, including boot, updates, identity, and communications.
- Classify by lifespan and data sensitivity. Devices that live fifteen years or protect long-lived secrets come first.
- Fix the root of trust. Where the first-stage boot code cannot be changed, add quantum-resistant verification before shipping.
- Plan for hybrid mode. Use classical plus post-quantum together during the transition, so you stay secure if either is broken.
- Budget the resources. Check flash, RAM, boot time, and message sizes with real post-quantum libraries on your target chip.
- Build crypto agility. Make sure update processes, including the ones described in our article on IoT device management, can replace algorithms across a fleet.
Common Challenges
| Challenge | Why it appears |
|---|---|
| Larger keys and signatures | They strain memory, bandwidth, and packet sizes on constrained networks |
| Performance cost | Slower operations affect boot time and battery life without hardware acceleration |
| Unchangeable boot code | Devices with fixed roots of trust cannot adopt new algorithms later |
| Moving guidance | Recommended algorithms and deadlines continue to be refined by regulators |
| Vendor claims | Certification scope varies, so confirm exactly what a certificate covers |
What Comes Next
Expect post-quantum support to become a standard line item on microcontroller and secure element datasheets, much as hardware AES acceleration did years ago. Expect procurement teams, especially in government-adjacent and critical infrastructure markets, to start asking for evidence of a transition plan. For a San Jose hardware team, the sensible stance is neither panic nor delay. Inventory the cryptography, protect the roots of trust that cannot be changed later, choose silicon that supports the new standards, and build the ability to update algorithms in the field. Devices designed that way will remain trustworthy for the long lives they are expected to have.
