Zero trust architecture: NIST SP 800-207, SP 800-207A, SP 1800-35 and CISA’s maturity model, explained
NIST SP 800-207 (August 2020) still defines zero trust as a set of tenets and three logical components, not a product. SP 800-207A, CISA’s five-pillar maturity model and the 2025 SP 1800-35 practice guide build on it, and CISA’s own dates for its model conflict.
The federal documents that define zero trust describe an architecture, not a product. NIST Special Publication 800-207, Zero Trust Architecture, published in August 2020, is still the final version on NIST’s publication page as of November 14, 2026 [1] [2]. Three later documents build on it: NIST SP 800-207A for cloud-native applications, CISA’s Zero Trust Maturity Model for planning progress, and NIST SP 1800-35, a practice guide with example builds [3] [4] [5]. None of them is a regulation; each is guidance. This explainer covers four parts: what SP 800-207 defines, how it says to migrate, what each later document adds, and where the documents disagree on dates.
What SP 800-207 defines
NIST describes zero trust as an evolving set of cybersecurity paradigms that move defenses from static, network-based perimeters to users, assets and resources [2]. Its central assumption is that no implicit trust is granted to assets or user accounts based only on their physical or network location or on who owns the asset [2]. Authentication and authorization of both subject and device are discrete functions performed before a session to an enterprise resource is established [2].
The publication sets out seven tenets [1]:
- all data sources and computing services are resources;
- all communication is secured regardless of network location;
- access to each resource is granted per session, with least privilege;
- access is decided by dynamic policy, including the state of the user’s identity, the application and the requesting device;
- the enterprise monitors and measures the integrity and security posture of all owned and associated assets;
- authentication and authorization are dynamic and strictly enforced before access is allowed; and
- the enterprise collects as much information as it can about assets, network and communications and uses it to improve its security posture [1].
NIST calls these tenets the ideal goal and acknowledges that not all may be fully implemented in their purest form [1].
The architecture rests on three logical components. The policy engine makes and logs the decision to grant, deny or revoke access; the policy administrator sets up or shuts down the communication path on that decision; and the policy enforcement point enables, monitors and ends the connection between a subject and a resource [1]. The publication describes three ways to build around them: enhanced identity governance, logical micro-segmentation and network-based segmentation, and says a full zero trust solution includes elements of all three [1].
For a platform team, the component model is the useful part. It separates the decision from the enforcement, which lets an organization keep existing gateways, identity systems and endpoint tools and ask a narrower question of each: does it make a policy decision, carry one out, or supply the data a decision needs.
How SP 800-207 says to migrate
NIST calls implementing zero trust a journey rather than a wholesale replacement of infrastructure, and expects most enterprises to run in a hybrid zero trust and perimeter-based mode for an indefinite period [1]. A greenfield build is, in NIST’s words, rarely a viable option for any organization with an existing network [1].
The migration steps in section 7.3 are to identify the actors on the enterprise, identify the assets the enterprise owns, identify key processes and evaluate their risks, formulate policies for a candidate workflow, identify candidate solutions, deploy and monitor, and then expand [1]. NIST says incomplete knowledge of assets, subjects and business processes will most often lead the policy engine to deny requests, and singles out unknown shadow IT as a source of that failure [1]. It suggests administrators may wish to run new policies in an observation and monitoring mode at first, because few policy sets are complete in their first iterations [1]. The steps can be mapped to the NIST Risk Management Framework, SP 800-37 [1].
In our view, that sequence is the best guide to sizing a program. An inventory of identities, devices and data flows comes first because the policy engine cannot decide without it, and NIST describes migration one business process at a time, starting from a candidate workflow [1].
What the later documents add
SP 800-207A (September 2023) applies zero trust to cloud-native applications in multi-cloud and hybrid environments [3]. Its abstract says the focus shifts from network parameters such as IP addresses and subnets to identities, including application and service identities, enforced through components such as API gateways, sidecar proxies and application identity infrastructure regardless of where the service runs [3].
CISA’s Zero Trust Maturity Model, version 2.0, is described by CISA as one of many roadmaps agencies can reference as they transition to a zero trust architecture [5]. It has five pillars, identity, devices, networks, applications and workloads, and data, and three cross-cutting capabilities, visibility and analytics, automation and orchestration, and governance [6]. Within each pillar, CISA’s web page says the model gives examples of traditional, initial, advanced and optimal zero trust architectures; the PDF calls Traditional the starting point and Initial, Advanced and Optimal the three stages of the journey [5] [6]. CISA says version 2.0 incorporates alignment to OMB Memorandum M-22-09, published in January 2022 [5]. The model says that, in alignment with NIST’s steps, agencies should assess their current enterprise systems, resources, infrastructure, personnel and processes before investing in zero trust capabilities [6]. It is written for federal agencies, but its pillars give any organization a way to report progress by area.
SP 1800-35 (June 2025) is a NIST cybersecurity practice guide that explains how to implement zero trust consistent with SP 800-207 [4]. NIST’s National Cybersecurity Center of Excellence worked with 24 collaborators under cooperative research and development agreements to build 19 example implementations from commercially available technology [4]. The guide includes technical detail for each build, lessons learned, and mappings of zero trust principles and technologies to commonly used security standards and guidelines [4].
Where the documents disagree on dates
One conflict is inside CISA’s model. The version 2.0 PDF is dated April 2023 on its cover and page headers, while its own revision history table lists version 2.0 as March 2022, described as the response to comments [6]. CISA’s web page says version 1.0 opened for public comment in September 2021, and the same revision table lists version 1.0 as August 2021 [5] [6]. We report the dates as each source gives them and do not resolve them.
Separately, CISA ties version 2.0 to M-22-09 [5]. We did not confirm the current status of M-22-09 against an official source on November 14, 2026, so this page does not say whether its deadlines still bind federal agencies.
Putting the documents together
For banks, health systems and agencies alike, the four documents answer different questions: SP 800-207 says what zero trust is, SP 800-207A applies it to cloud-native services, the maturity model gives a way to measure progress by pillar, and SP 1800-35 shows worked builds [1] [3] [4] [6]. In our view, an architecture review that names which component plays the policy engine, administrator and enforcement point for each critical workflow gives a clearer picture than a list of tools bought. How any of these documents applies to a given organization depends on its regulators and contracts, which organizations should confirm with counsel. This explainer is not legal advice.
Moraph’s teams help organizations with work like this. See our solutions.
Spotted an error? Tell us and we will check it and correct the explainer where needed.