What Is SBOM Management? A Practical Guide for Device Makers
It's never been more important to have a strong SBOM management strategy.

Finite State Team
Most teams can produce an SBOM. Fewer can produce the right one on the day a customer, an auditor, or ENISA asks for it. The distance between those two is SBOM management.
Below is what SBOM management covers, how it differs from dependency scanning, the lifecycle that keeps an SBOM current, where it breaks on connected devices, and what the 2026 rules expect from it.
What is SBOM management?
SBOM management is the practice of generating, collecting, verifying, enriching, and sharing software bills of materials across every release, so the inventory you hand over matches the software you shipped.
An SBOM on its own is a snapshot. It describes one build of one product at the moment someone ran a tool. Ship a patch, swap a supplier library, or rebuild with a new toolchain, and that file stops describing the product.
Management is what keeps the record true. It covers your own SBOMs and the ones suppliers send you, the vulnerability data attached to every component, and the decisions your team makes about each finding.
It also decides who can use the SBOM. Engineering plans upgrades from it. The PSIRT answers "are we affected?" from it. Compliance hands it over as evidence. One record serves all three, or none of them trust it.
SBOM management vs dependency scanning
Dependency scanning answers a question about code at one moment. SBOM management answers it for the product across its life.
| Dependency scanning | SBOM management | |
|---|---|---|
| Unit of work | A repository or build at one point in time | A product version, tracked across releases |
| Reads | Manifests and lock files | Manifests, supplier SBOMs, and the shipped binary |
| Supplier components | Only what your build declares | Supplier SBOMs ingested and checked against the build |
| Output | A list of findings | An inventory with findings, decisions, and history attached |
| When a new CVE lands | Re-scan the code that is still open | Check every shipped version already in the inventory |
| Triage decisions | Live in tickets, or nowhere | Recorded as VEX and carried forward |
| Audience | Developers | Engineering, PSIRT, compliance, and customers |
The two are not rivals. Scanning is a good input to SBOM management, especially early in development.
The difference shows up after release. A scanner watches the code you are still working on. The products in the field need someone watching what they actually run.
The SBOM management lifecycle
Five stages, each producing an output someone downstream depends on.
- Generate an SBOM for every release. If the software changes, you have a new SBOM. Generate from source or the build for developer feedback, and from the shipped image for the record of what went out. Generators handle those inputs differently, which is the main thing to test when comparing SBOM generation tools.
- Ingest and verify supplier SBOMs. Collect the SBOMs your chipset vendors, ODMs, and software suppliers send, normalize them to one format, and check them against what the delivered binary contains. A supplier SBOM is a claim until something confirms it.
- Enrich and monitor. Attach vulnerability, exploit, license, and end-of-support data to every component, then keep matching as that data changes. An unchanged SBOM produces new findings every week, because the databases underneath it move. SBOM vulnerability management covers the matching and triage loop in detail.
- Record decisions as VEX. For each finding, state whether the product is affected and why, in a machine-readable form. Every recorded decision is one your team never makes twice.
- Share and retain. Decide what goes to customers, auditors, and regulators, and what stays internal. Version SBOMs alongside releases and keep them for as long as the product is supported.

Stage four is where most programs stall. Teams generate SBOMs and even monitor them, then triage findings in spreadsheets that never travel with the SBOM. The next scan raises the same findings, and the work starts over.
Where SBOM management breaks on connected devices

Most SBOM management advice assumes a web application with a package manager. A connected device breaks that assumption in four places.
- Firmware has no manifest. Statically linked libraries, RTOS images, and cross-compiled binaries leave no package record for a manifest-driven tool to read. They generate no findings, which reads as a clean scan.
- Supplier code arrives compiled. Board support packages and ODM blobs come as binaries, often with no SBOM and sometimes with no version string.
- Products outlive their components. A device that stays in service for 10 years will run libraries their maintainers stopped supporting. End-of-support dates belong in the inventory, not in someone's memory.
- Many versions ship at once. A manufacturer may support dozens of SKUs and firmware versions in parallel. A new CVE has to be checked against every one of them, not just the current release.
Your build manifest records what you intended to ship. The binary records what shipped. For a device program, SBOM management means reconciling the two, which requires analyzing the shipped image rather than trusting the declaration.
SBOM management compliance requirements in 2026

Four rules define what a managed SBOM has to hold in 2026. None of them names a tool, and all of them assume the inventory already exists.
| Rule | What it asks of your SBOM program | Status |
|---|---|---|
| CISA 2026 Minimum Elements | 17 data fields, up from seven, including component hash, license, and generation context; all components including transitive dependencies, with no minimum depth | Published 29 July 2026; replaces the 2021 NTIA baseline; guidance, not law |
| EU CRA Article 14 | Report actively exploited vulnerabilities: early warning within 24 hours, notification within 72 hours, final report within 14 days of a corrective measure | Applies since 11 September 2026, including products already on the market |
| EU CRA Annex I and Annex VII | An SBOM in a commonly used, machine-readable format covering at least top-level dependencies, kept in the technical documentation | Applies from 11 December 2027 |
| FDA Section 524B(b)(3) | An SBOM covering commercial, open-source, and off-the-shelf components; current guidance also asks for each component's support status and end-of-support date | In force since 29 March 2023; guidance revised 3 February 2026 |
Article 14 tests SBOM management hardest. Twenty-four hours leaves no time to find out what is inside a product. You either know which shipped versions contain the exploited component, or you spend the first day finding out. The CRA Article 14 reporting deadline post covers the trigger and the workflow.
For medical devices, the FDA's support-status fields are where generic inventories fall short. Our breakdown of FDA SBOM requirements for medical devices walks through each one.
Note the gap between baselines. The CRA's legal floor is top-level dependencies, while the CISA elements ask for all of them. A manufacturer meeting only the floor is compliant and will still receive customer questionnaires written against the higher bar.
SBOM management best practices
- Scope by obligation, then plan past it. Start with the products a regulator or customer asks about, but plan for full dependency depth from the beginning.
- Generate on every release, automatically. Tie generation to the build or release pipeline, so no one has to remember.
- Generate from the shipped artifact as well as the source. A pre-build SBOM arrives earlier. An after-build SBOM can carry hashes and catch what the manifest missed. Keep both and reconcile them.
- Treat supplier SBOMs as claims. Verify them against the delivered binary before they enter your inventory.
- Name one owner. Product security, the PSIRT, or compliance can own it, but one team does, with named handoffs to the others.
- Record every triage decision as VEX. A not-affected call with a written reason is evidence. The same call in a chat thread is not.
- Decide what you share before anyone asks. Define the customer-facing SBOM and the internal one in advance, so a request takes minutes rather than a legal review.
What to look for in an SBOM management tool
Put these questions to any vendor, including us. The answers separate a file generator from a management program.
- Can it generate an SBOM from a compiled firmware image with no source code?
- Can it ingest supplier SPDX and CycloneDX files and show where they disagree with the binary?
- Does it populate the 2026 CISA fields, including component hash and generation context, or mark them unknown?
- When a new CVE lands, does it check every shipped version already in the inventory without a re-scan?
- Can it record VEX decisions with the reasoning attached and carry them to the next release?
- Does it track support status and end-of-support dates per component?
- Can it export what a regulator asks for: SPDX, CycloneDX, VEX, and a report mapped to the specific rule?
- Does it hold up at 100 products, not just one?
Key takeaways
- An SBOM is a snapshot. SBOM management keeps it true across releases, suppliers, and new vulnerability data.
- Scanning and management answer different questions. One checks code at a moment; the other tracks a product across its life.
- Supplier SBOMs are claims. Check them against the binary before they enter your inventory.
- Triage decisions are the asset. Recorded as VEX, they stop the same findings from resetting every quarter.
- The 2026 rules assume the inventory exists. CRA Article 14 gives you 24 hours, which is no time to build it.
Where to start
One first move. Pick your highest-exposure shipped product and ask how long it would take to say whether a given component is present in every version still in the field. If the answer is longer than a day, Article 14 has already found your gap.
Finite State generates SBOMs from firmware and binaries with no source code required, reconciles supplier SBOMs against what the image contains, and re-evaluates VEX decisions as software and exposure change. See how that works on our SBOM lifecycle management page, or request a demo against one of your own builds.
FAQ
What is SBOM management?
SBOM management is the practice of generating, collecting, verifying, enriching, and sharing software bills of materials across every release. It keeps the inventory you hand to customers and regulators consistent with the software you actually shipped, including supplier components and the vulnerability decisions attached to them.
How do you create and maintain an SBOM?
Generate one for every release, from the build and from the shipped artifact, and ingest supplier SBOMs alongside it. Then monitor every component against vulnerability and end-of-support data, record triage decisions as VEX, and version the SBOM with the release it describes.
What is the difference between SBOM management and dependency scanning?
Dependency scanning checks a repository or build for known vulnerabilities at one point in time. SBOM management tracks a product's full inventory across releases, including supplier and binary-only components, and keeps the findings and decisions attached to each shipped version.
What are the SBOM management compliance requirements?
The EU CRA requires 24-hour reporting of actively exploited vulnerabilities since September 2026, and an SBOM in the technical documentation from December 2027. FDA Section 524B requires an SBOM in premarket submissions for cyber devices. The CISA 2026 Minimum Elements set the data baseline customers increasingly ask for.
Should we use SPDX or CycloneDX?
Either. Both satisfy the CISA 2026 baseline, and most generators emit both. Choose the one your tools already produce, record the format and version in the SBOM itself, and expect some customers to ask for the other.
How often should an SBOM be updated?
Generate a new SBOM whenever the software changes, which in practice means every release. Monitoring is continuous: the vulnerability data underneath an unchanged SBOM changes daily, so its findings do too.

Finite State Team
The Finite State team brings together experts in cybersecurity, embedded systems, and software supply chain risk to help connected device manufacturers secure their products and comply with evolving global regulations.