NVD backlog and CVE volume, 2016 to 2026: records double, enrichment narrows

We counted every CVE in NIST's National Vulnerability Database by year and status to see how fast volume grew and how much of it NIST still enriches.

Why the NVD matters to every patching program

NIST's National Vulnerability Database (NVD) takes every published CVE record and, when it enriches one, adds reference tags, a CVSS severity score, a CWE weakness type and CPE applicability statements, the machine-readable list of affected products [8]. CVE records usually appear in the NVD within an hour of publication to the CVE List; enrichment is the slower, manual step [8].

Until 2024 that step kept pace. In February 2024 NIST warned of temporary delays in analysis [6], and since then it has issued a run of updates on a backlog it has not cleared. In April 2026 it stopped trying to enrich every CVE [4]. We counted what the NVD holds today, year by year.

Data and method

Dataset: the NVD CVE API 2.0 at https://services.nvd.nist.gov/rest/json/cves/2.0, queried on November 16, 2026 without an API key [1]. The API's published-date filter requires both pubStartDate and pubEndDate and allows at most 120 consecutive days per request, and it returns at most 2,000 records per page [2]. We therefore split each calendar year from 2016 to 2026 into quarters (the last one ending November 15, 2026) and paged through every record with requests of this form [1]:

  • GET https://services.nvd.nist.gov/rest/json/cves/2.0?pubStartDate=2024-01-01T00:00:00.000&pubEndDate=2024-03-31T23:59:59.999&resultsPerPage=2000&startIndex=0, then startIndex 2000, 4000 and so on until it reached totalResults [1].
  • Windows used each year: 01-01 to 03-31, 04-01 to 06-30, 07-01 to 09-30 and 10-01 to 12-31, each from T00:00:00.000 to T23:59:59.999 with no offset, so in GMT per the API's default; for 2026 the last window was 2026-10-01 to 2026-11-15 [2].
  • The public rate limit without a key is 5 requests in a rolling 30-second window [3]; we waited at least 6.5 seconds between requests and retried after any error, so 44 windows took about two hours [1].

For each record we kept the CVE ID, the published timestamp, vulnStatus, whether any CVSS metric lists nvd@nist.gov as its source, and whether the record carries configurations (CPE applicability statements) [1]. Every window's record count matched its totalResults, 351,542 records in all [1]. "Published" is the date the CVE was added to the NVD [2]. Yearly totals include rejected records; status shares use non-rejected records only. "Enriched" means vulnStatus Analyzed or Modified; "not enriched" means Deferred, Awaiting Analysis, Received or Undergoing Analysis. The API's Deferred, Awaiting Analysis and Analyzed are shown on the NVD website as Not Scheduled, Awaiting Enrichment and Enriched [7]. In April 2026 NIST moved the CVEs it had Deferred in 2025 to Modified After Enrichment [4], so Modified can include some records NIST never fully enriched. Statuses are as of November 16, 2026 and will change as NIST works; percentages are rounded to one decimal place.

Volume: a steady climb, then a break in 2026

CVEs in the NVD by published year, including rejected records, as queried on November 16, 2026
Year publishedCVEsOf which rejected
20166,517 [1]68 [1]
201718,113 [1]3,471 [1]
201818,154 [1]1,644 [1]
201918,938 [1]1,633 [1]
202019,222 [1]900 [1]
202121,950 [1]1,801 [1]
202226,431 [1]1,358 [1]
202330,949 [1]2,133 [1]
202440,704 [1]751 [1]
202549,972 [1]1,822 [1]
2026 (to November 15)100,592 [1]2,773 [1]

From 2017 to 2020 the NVD published between 18,000 and 20,000 CVEs a year; the count then rose by 32% in 2024 and 23% in 2025 [1]. NIST's own figures run the same way: it reported that CVE submissions rose 32 percent in 2024 [4] and 263% between 2020 and 2025 [5], against 160% growth in the records we counted by NVD published date over the same years [1]. The two figures measure different things: NIST's 263% is its count of CVE submissions, a measure it does not define on that page, while ours counts records by the date they were added to the NVD, including rejected records, as the API returns them today, so the two need not match [5] [2].

2026 is a break in the series. Between January 1 and November 15, 2026 the NVD published 100,592 CVEs against 43,063 in the same span of 2025, a rise of 134% [1]. The first quarter was 31% above the first quarter of 2025, in line with NIST's statement that early-2026 submissions were nearly one-third higher [1] [5]. The surge came later: monthly publications ranged from 3,116 to 5,643 through 2025, then climbed from 5,143 in January 2026 to 9,919 in July and 17,481 in October [1]. On November 16, 2026 the NVD Dashboard showed 100,666 new CVEs received by the NVD this year, a figure that includes that day's records, close to our count [9].

Moraph's view

Moraph's view: the records we pulled do not say who filed the extra CVEs or why, so we do not attribute the 2026 surge to any one cause. Whatever drove it, a doubling in one year is far beyond what a manual enrichment team can absorb, and a large part of the growth arrived after NIST had already narrowed what it enriches.

The backlog: how enrichment fell behind

NIST's public account of the backlog, in its own updates, runs as follows. On February 13, 2024 it announced temporary delays in analysis while it worked to set up a consortium [6]. On April 25, 2024 it reported a growing backlog, citing more software and therefore more vulnerabilities, and a change in interagency support [6]. On May 29, 2024 it said it had awarded a contract for extra processing support and expected the backlog to be cleared by the end of the fiscal year [6]. On November 13, 2024 it said that estimate had been optimistic, although all KEV entries in the backlog had been addressed [6]. On March 19, 2025 it said the backlog was still growing [4].

On April 2, 2025 NIST said that CVEs published before January 1, 2018 and still awaiting enrichment would be marked Deferred, meaning it did not plan to prioritize them because of their age [4]. On April 15, 2026 it went further [4] [5]:

  • Enrichment is now prioritized for CVEs in CISA's KEV catalog, with a goal of one business day; CVEs for software used within the federal government; and CVEs for critical software as defined by Executive Order 14028 [4].
  • Other CVEs are still added but marked lowest priority and not scheduled for immediate enrichment; anyone can ask NIST by email to enrich one [4].
  • All backlogged CVEs published before March 1, 2026 moved to Not Scheduled, and the CVEs Deferred in 2025 moved to Modified After Enrichment [4].
  • NIST no longer routinely adds its own severity score when the CVE Numbering Authority supplied one, and re-analyzes modified CVEs only when a change materially affects its enrichment [4].

NIST said it enriched nearly 42,000 CVEs in 2025, 45% more than any prior year, and that this was still not enough [5]. On November 16, 2026 the NVD Dashboard's CVE Status Count showed 63,104 CVEs as Deferred, 11,626 as Awaiting Analysis and 5,159 as Received [9]; the last two change through the day. Our records match the Deferred figure exactly: all 63,104 Deferred CVEs were published in 2024, 2025 or 2026, and none of our records published from 2016 to 2023 is Deferred [1].

Status by year: what is enriched today

Status of non-rejected CVEs by published year, as queried on November 16, 2026
Year publishedNon-rejected CVEsEnriched (Analyzed or Modified)Deferred (Not Scheduled)Awaiting, Received or UndergoingWith CPE configurationsWith an NVD-sourced CVSS score
202328,816 [1]100.0% [1]0.0% [1]0.0% [1]100.0% [1]92.9% [1]
202439,953 [1]80.5% [1]19.4% [1]0.0% [1]80.6% [1]58.3% [1]
202548,150 [1]62.7% [1]37.3% [1]0.0% [1]62.7% [1]34.8% [1]
2026 (to November 15)97,819 [1]43.2% [1]38.2% [1]18.6% [1]43.4% [1]14.9% [1]

All but 15 of the non-rejected CVEs published from 2016 to 2023 are in Analyzed or Modified status today, and every one carries CPE configurations [1]. From 2024 the picture changes: 7,762 CVEs published in 2024, 17,954 from 2025 and 37,388 from 2026 so far are Deferred [1]. Counting the 18,201 CVEs from 2026 still awaiting, received or undergoing analysis, 55,589 of this year's 97,819 non-rejected CVEs, or 56.8%, have no NVD enrichment yet [1].

The share enriched falls with the month of publication, partly because recent CVEs have had less time in the queue: 71.3% of non-rejected CVEs published in March 2026 are enriched, against 27.5% of those from October and 10.9% of those from November 1 to 15 [1]. Of CVEs published since April 15, 2026, 39.9% are already Deferred [1]. The drop in NVD-sourced severity scores, from 92.9% of 2023 CVEs to 14.9% of 2026 CVEs, mostly reflects the policy change: NIST now leaves the severity score to the numbering authority when one is supplied [1] [4].

Moraph's view

Moraph's view: the NVD has moved from a database that enriched everything, slowly, to one that enriches a defined slice quickly and lists the rest. For federal agencies the slice is close to what they run. For banks and health systems, much of their software (clinical systems, core banking platforms, niche vendor appliances) may fall outside it, so the gap between "a CVE exists" and "my scanner knows it affects me" now depends on where that product data comes from, not on NIST.

What fills the gap: CISA's Vulnrichment and the CVE record itself

CISA announced Vulnrichment on May 10, 2024, adding Stakeholder-Specific Vulnerability Categorization (SSVC) decision points, CWE identifiers and CVSS scores to CVE records through its Authorized Data Publisher (ADP) container [10]. CISA scores each CVE it assesses on three SSVC decision points (exploitation, automatable and technical impact) and adds CWE or CVSS data mainly for higher-risk CVEs where there is enough evidence; it never overwrites the numbering authority's own data [11]. Since July 3, 2024 the NVD has shown Vulnrichment CVSS and CWE data on its detail pages, and since June 17, 2026 it has carried SSVC data and the CVE record's own "affected" product list in its API [6] [4].

These additions help with prioritization but do not replace NVD's CPE applicability statements, which are missing for more than half of this year's CVEs [1].

What the data do not show

The API gives each CVE's current status, not its history, so these counts cannot say how long a given CVE waited; a later query will show different shares as NIST enriches on request or by priority. The records do not say whether a Deferred CVE affects federal or critical software, so we cannot measure how well NIST's new priority criteria are applied. Published dates are when the NVD added a record, which can differ from when a vendor disclosed the flaw [2]. CVE counts also depend on how numbering authorities split and file flaws, so more CVEs is not by itself more risk.

Moraph's view

Moraph's view: plan for the NVD as it now is, not as it was. Expect continued high CVE volume through 2027, a growing Not Scheduled pile and NIST enrichment concentrated on KEV, federal and critical software. A vulnerability program that matches CVEs to assets should be able to work from the CVE record's own affected data, CISA's ADP fields and vendor advisories when NVD CPE data are absent, and should track how many of its open findings depend on CVEs that NIST has not enriched.

What this means for you

  • Vulnerability management leads: Measure how many open findings in your environment rely on CVEs without NVD CPE configurations, as 56.6% of non-rejected 2026 CVEs were on November 16, 2026 [1], and add the CVE record's affected data, CISA's ADP fields and vendor advisories as matching sources where NVD data are absent [4] [11].
  • CISOs at banks and health systems: Ask scanning and asset-management suppliers how they detect CVEs that NIST has marked Not Scheduled; NIST's enrichment scope now covers KEV, federal and EO 14028 critical software [8], which may not include clinical or core banking products.
  • Federal agencies and their contractors: Expect fast NVD enrichment for KEV entries, with a one-business-day goal [4], and for software used within the federal government [8]; request enrichment by email for a lowest-priority CVE that matters to you, as NIST invites [4].
  • Risk and audit teams: Stop treating the NVD severity score as universal: only 14.9% of non-rejected 2026 CVEs carry a score from NIST, so any score shown is usually the numbering authority's or CISA's, if there is one [1] [4]. Policies that set patch deadlines by 'NVD CVSS' should say which score counts when NIST provides none.

Moraph predictions

Every prediction is Moraph's view, not a fact. Each has a date, a confidence level (70% means we expect to be wrong about three times in ten) and a test anyone can apply. We score them on this page when the date comes and never remove one. How we make and score predictions.

  1. P1 By Confidence 70% Open

    The NVD lists at least 120,000 CVEs (including rejected records) with a published date in calendar year 2027.

    How we will check: On or after January 31, 2028, query the NVD CVE API 2.0 with pubStartDate and pubEndDate windows of at most 120 days covering 2027-01-01T00:00:00.000 to 2027-12-31T23:59:59.999 and add the totalResults values. Right if the sum is 120,000 or more.

    Basis: The NVD published 100,592 CVEs from January 1 to November 15, 2026, including 17,481 in October alone, against 49,972 in all of 2025 [1]; NIST said it does not expect the surge in submissions to let up [5].

  2. P2 By Confidence 75% Open

    Fewer than half of the non-rejected CVEs published to the NVD in calendar year 2027 are in Analyzed or Modified status on January 31, 2028.

    How we will check: On January 31, 2028 (or the first day after it that the API answers), page through all CVEs with a 2027 published date as in this report's method, drop vulnStatus Rejected, and divide the Analyzed plus Modified count by the rest. Right if under 50%.

    Basis: 43.2% of non-rejected 2026 CVEs were enriched as of November 16, 2026, and 27.5% of those published in October 2026 [1]; NIST now enriches only CVEs that meet its priority criteria [4].

  3. P3 By Confidence 60% Open

    The NVD Dashboard's Deferred count reaches 120,000 or more on or before December 31, 2027.

    How we will check: Read the CVE Status Count table at https://nvd.nist.gov/general/nvd-dashboard on December 31, 2027 (or an archived copy of that date). Right if the Deferred row shows 120,000 or more.

    Basis: The dashboard showed 63,104 Deferred CVEs on November 16, 2026 [9]; 31,759 CVEs published since April 15, 2026 were already Deferred as queried that day [1]. Reaching 120,000 needs about 4,200 more a month, and enrichment on request can move some out of Deferred [4].

  4. P4 By Confidence 80% Open

    On December 31, 2027 the NVD's published scope of coverage still limits enrichment to named categories of CVEs rather than all CVEs.

    How we will check: Read the 'NVD Scope of Coverage' section of https://nvd.nist.gov/general/cve-process (or an archived copy dated on or near December 31, 2027). Right if it still lists specific categories (such as KEV, federal or critical software) instead of stating that all CVEs are enriched.

    Basis: The page limits scope to KEV entries, software used within the federal government and EO 14028 critical software [8]; NIST announced the change on April 15, 2026, calling it necessary to manage the current surge in CVE submissions and saying it frees resources to develop automated systems [4].

Facts on this page link to their sources. Text marked "Moraph's view", and every prediction, is our own analysis, not a fact or advice. Methodology.

Spotted an error? Tell us and we will check it and correct the piece where needed.

  • NIST
  • NVD
  • CVE
  • vulnerability management
  • CISA
  • Vulnrichment
  • KEV
  • patching