PharmaDossier
Manufacturing

FDA device clearances & approvals by the numbers (2026)

FDA database analysis of medical device throughput: 510(k) clearances, PMA originals vs supplements, De Novo pathways, and the record surge in AI/ML software clearances.

Ran Chen
Ran Chen
20 min read · Published · Source-cited

When medical device manufacturers, regulatory affairs (RA) professionals, and business development leaders assess the U.S. market, they are often confronted with a complex, multi-tiered regulatory authorization landscape. Unlike drugs, which are typically approved via standard New Drug Applications (NDAs) or Biologics License Applications (BLAs), medical devices enter the U.S. market through several distinct regulatory gates. The three primary premarket pathways are Premarket Notification [510(k)], Premarket Approval (PMA), and the De Novo classification pathway.

For a medtech or combination-product team sizing the FDA device-authorization landscape for 2026 planning, the numbers reveal a clear picture of CDRH throughput. The FDA authorizes roughly 3,000 to 3,300 devices per year via the 510(k) pathway (the database shows 3,225 cleared in calendar year 2025, matching the FDA's reported 3,238 clearances in fiscal year 2025). In stark contrast, the FDA grants only 30 to 40 first-time (original) PMA approvals per year for high-risk Class III devices. Across the FDA's historical medical device database, there are 175,456 cumulative 510(k) records and 56,683 PMA records. However, a critical inspection of the PMA base reveals that 97.4% of all PMA records (55,213) are post-market supplements, while only 2.6% (1,470) represent original approvals—resulting in a ratio of 37.6 supplements filed for every single original PMA.

From a panel perspective, cardiovascular devices dominate both major pathways, accounting for 19,084 cumulative 510(k) clearances (10.9% of the cohort) and 28,771 PMA records (including 448 originals). The most significant growth vector in device authorizations is the surge in artificial intelligence and machine learning (AI/ML)-enabled software. The FDA's official list hit a record 295 AI/ML clearances in calendar year 2025 (up from 253 in 2024 and 221 in 2023), with approximately 94% to 95% routing through the 510(k) pathway. Meanwhile, the De Novo pathway for novel moderate-risk devices has slowed to a crawl, registering only 3 grants in the first quarter of 2025 with dozens of applications pending, as manufacturers adjust to the mandatory electronic submission template (eSTAR) requirement that took effect on October 1, 2025. Additionally, the FDA's Breakthrough Devices Program has granted 1,284 designations as of March 31, 2026, creating a major pipeline of priority-review candidates.


How many devices does FDA clear/approve each year, by pathway?

Evaluating the regulatory throughput of the FDA's Center for Devices and Radiological Health (CDRH) requires a deep dive into annual decision counts. The vast majority of medical devices enter the market via the 510(k) premarket notification pathway, which is designated for moderate-risk Class II devices (and some low-risk Class I devices) that can demonstrate "substantial equivalence" to a legally marketed predicate device. The PMA pathway is reserved for high-risk Class III devices that support or sustain human life, are of substantial importance in preventing impairment of human health, or present a potential unreasonable risk of illness or injury.

To illustrate the volume discrepancy between these pathways, the table below compiles annual decision data from the openFDA registry, comparing calendar-year 510(k) clearances against total PMA decisions and original (first-time) PMA approvals.

Annual Regulatory Decisions by Premarket Pathway (2015–2025)

Calendar Year 510(k) Clearances Total PMA Decisions (Original + Supplement) Original PMA Approvals (Class III) PMA Originals Share of Total (%)
2015 3,020 2,169 43 1.98%
2016 2,956 2,564 39 1.52%
2017 3,200 2,717 46 1.69%
2018 3,062 2,790 31 1.11%
2019 2,918 2,638 32 1.21%
2020 2,923 2,503 34 1.36%
2021 3,023 2,204 31 1.41%
2022 3,209 2,092 22 1.05%
2023 3,346 2,185 35 1.60%
2024 3,129 2,230 33 1.48%
2025 3,225 2,222 40 1.80%
2026 (H1) 1,661 1,157 23 1.99%
Cumulative 175,456 56,683 1,470 2.59%

Data Source: openFDA Device Database (as-of snapshot July 6, 2026). Cumulative totals reflect all records logged in the database since inception.

These figures demonstrate a consistent operational baseline. The FDA clears between 3,000 and 3,300 510(k) submissions annually, while original PMA approvals consistently hover in the narrow band of 30 to 45 per year. In terms of overall market entry, new Class III PMA approvals account for less than 1.5% of the combined premarket clearance and approval volume.

The 510(k) pathway is further divided into three main submission types:

  1. Traditional 510(k): The standard method, accounting for 156,275 records (89.1%) of the historical base. It requires a complete comparison to a predicate device and can take several months to review.
  2. Special 510(k): Utilized for device modifications made by the original manufacturer that do not alter the intended use or fundamental scientific technology. This streamlined route represents 16,034 records (9.1%) of all clearances.
  3. Abbreviated 510(k): Relies on FDA guidance documents, special controls, or recognized consensus standards rather than a direct comparison to a specific predicate. This type represents 2,593 records (1.5%) of the registry.

This pathway division shows that the overwhelming majority of 510(k) submissions are Traditional, indicating that manufacturers continue to prioritize full predicate comparisons to establish equivalence. For teams managing active product lifecycles, understanding these baselines is critical when scheduling regulatory submissions and planning market entry timelines. To contextualize how these premarket numbers compare to post-market safety and enforcement actions, refer to our analysis on drug-delivery device recalls by the numbers: infusion pumps.


Why are 97% of PMA records supplements, and what do those supplements change?

One of the most common misconceptions in medtech and biopharma business intelligence is the belief that every "PMA approval" logged by the FDA represents a new Class III medical device entering the market. In reality, the premarket approval database is overwhelmingly composed of post-market modifications. Out of 56,683 total PMA records in the FDA registry, 55,213 (97.4%) are supplements, while only 1,470 (2.6%) are original PMAs.

This means that for every new, first-of-its-kind Class III medical device approved by the FDA, the manufacturer will subsequently file an average of 37.6 supplements over the life of that device family. This volume highlights the extensive regulatory change-control requirements associated with Class III hardware and software. Any modification that affects the safety or effectiveness of a PMA-approved device—including changes in manufacturing processes, sterilization sites, components, software algorithms, materials, and clinical indications—requires a formal supplement submission to CDRH.

To understand what these supplements actually modify, we can aggregate the PMA database by the documented supplement_reason. The table below outlines the primary categories of Class III modifications:

Breakdown of FDA PMA Supplements by Regulatory Reason

Rank Supplement Reason / Category Historical Count Share of Total Supplements (%) Key Regulatory Implications
1 Process Change (Manufacturer, Sterilizer, Packager, Equipment) 30,649 55.51% Changes in manufacturing location, scaling equipment, shifting sterilization contractors, or modifying quality control testing methods.
2 Change Design, Components, Specifications, or Materials 11,556 20.93% Modifying the physical device design, replacing electronic components, updating polymers or metals, or changing internal software code.
3 Labeling Change (Indications, Instructions, Shelf Life, Warnings) 4,709 8.53% Expanding clinical indications, modifying patient labeling, updating instructions for use (IFU), or extending product expiration dates.
4 Location Change (Manufacturer, Sterilizer, or Packager Site Shift) 3,157 5.72% Relocating primary manufacturing operations or shifting to a new contract manufacturing organization (CMO) facility.
5 Post-Approval Study (PAS) Protocol Updates 709 1.28% Modifying the protocols or reporting structures for FDA-mandated post-market clinical studies.
6 Post-Approval Study Protocol - Office of Surveillance & Biometrics 494 0.90% Specific surveillance protocol changes managed by the CDRH Office of Surveillance and Biometrics.
7 Express GMP Supplement 202 0.37% Streamlined manufacturing changes under Good Manufacturing Practice (GMP) guidelines that do not require pre-approval inspects.
8 Miscellaneous / Other Reasons 3,737 6.76% Unspecified administrative changes, ownership transfers, or minor compound updates.

Data Source: openFDA PMA Database (July 6, 2026 snapshot). Note that the raw data includes 4,456 blank/uncategorized entries that have been excluded from the percentage calculation.

This breakdown reveals that over half of all Class III device filings (55.5%) are driven by manufacturing process changes, and another 20.9% are driven by design and material updates. For drug-device combination products—such as prefilled syringes, autoinjectors, and drug-eluting stents—this change-control structure creates significant operational complexity. A biopharma company using an approved autoinjector platform must track every component change or manufacturing site shift made by their device partner, as these modifications can trigger secondary FDA supplement filings for the combination product. For more details on navigating these regulatory interfaces, see our guide on the FDA drug-device combination product regulatory pathway.

Furthermore, this supplement volume highlights the regulatory burden of "change control mismatch" between different global jurisdictions. While the FDA utilizes these tiered supplement categories (such as Real-Time Reviews, 180-Day Supplements, and 30-Day Notices) to process modifications, other regions like the European Union under the Medical Device Regulation (MDR) or individual national agencies in Asia may require full de novo applications for the same manufacturing changes, creating significant global supply chain friction.


Which advisory-committee panels and applicants dominate device authorizations?

Just as drug approvals are concentrated within specific therapeutic areas like oncology and immunology, medical device authorizations are highly concentrated within specific clinical specialties and advisory-committee panels. By analyzing the medical specialty codes in 510(k) clearances and the advisory committee panels in PMA records, we can identify which clinical areas command the most regulatory activity.

For 510(k) clearances, the top five medical specialties represent 47.1% of all historical clearances:

  1. Cardiovascular: 19,084 clearances (10.9%)
  2. Orthopedic: 16,994 clearances (9.7%)
  3. General and Plastic Surgery: 16,230 clearances (9.3%)
  4. General Hospital: 15,601 clearances (8.9%)
  5. Radiology: 14,585 clearances (8.3%)

In the PMA pathway, the concentration is even more extreme. The Cardiovascular panel accounts for 28,771 records (50.8% of the entire PMA database), including 448 original PMA approvals. This massive footprint is driven by the high risk and therapeutic importance of cardiac devices, such as pacemakers, implantable cardioverter-defibrillators (ICDs), transcatheter heart valves, coronary stents, and active circulatory support systems. The second-largest PMA panel is Ophthalmic with 6,541 records (202 originals), driven by contact lenses, intraocular lenses (IOLs), and refractive lasers, followed by General and Plastic Surgery with 3,101 records (89 originals), which includes dermal fillers and breast implants.

Applicant Leaderboards: Who holds the most original PMAs?

To identify the market leaders in high-risk Class III innovation, we can aggregate original PMA approvals (excluding supplements) by applicant name. Because the raw FDA database contains multiple legal-entity variants for the same corporate parent, we consolidate these listings to build an accurate corporate leaderboard.

Top Original PMA Applicants (First-Time Class III Approvals)

Corporate Parent (Consolidated) Original PMA Count Key Approved Technologies Clinical Specialties
Boston Scientific 75 Coronary stents, cardiac mapping, transcatheter valves, defibrillators. Cardiovascular, Electrophysiology, Urology
Abbott Laboratories / Abbott Medical 56 Continuous glucose monitors (CGM), transcatheter mitral valve repair, pacemakers. Cardiovascular, Neuromodulation, Diabetes Care
Medtronic 41 Implantable pacemakers, neurostimulators, insulin pumps, heart valves. Cardiovascular, Neurology, Orthopedics, Diabetes
Bausch & Lomb 22 Contact lenses, ophthalmic surgical systems, intraocular lenses. Ophthalmic
DiaSorin 19 High-complexity in vitro diagnostic (IVD) assays and instruments. Clinical Chemistry, Microbiology
Edwards Lifesciences 18 Transcatheter aortic valve replacement (TAVR), surgical heart valves. Cardiovascular
Hologic 18 Digital mammography systems, cervical cancer screening assays. Radiology, Obstetrics & Gynecology

Data Source: openFDA PMA Database (July 6, 2026 snapshot). Consolidated counts combine legal-entity variants (e.g., "Boston Scientific Corp," "Boston Scientific Corporation," and "Boston Scientific").

This applicant leaderboard is dominated by the "big three" of medtech: Boston Scientific, Abbott, and Medtronic. These firms have established deep regulatory expertise and capital reserves required to navigate the intensive clinical trial demands of original PMA submissions. In contrast, smaller medtech startups rarely seek original PMAs directly; instead, they focus on the De Novo or 510(k) pathways, or position themselves as acquisition targets for larger firms that can absorb the PMA post-market compliance costs.

From a geographic perspective, 510(k) applicant countries show a strong domestic concentration, with the United States representing 146,433 clearances (83.5%) of the historical base. However, foreign applicants are establishing a significant foothold, led by China at 5,579 clearances (3.2%), followed by Germany at 2,446 (1.4%), Canada at 2,235 (1.3%), and South Korea at 2,021 (1.2%). This geographic distribution is shifting rapidly, as foreign manufacturing sites expand their capabilities in Class II electronic devices and diagnostics. To explore how these premarket applicant concentrations translate into real-world patient safety reports, see our companion analysis on MAUDE drug-delivery device adverse events by the numbers.


How big is the AI/ML-enabled device surge, and which pathway do those devices use?

While traditional hardware segments like orthopedics and cardiovascular show stable regulatory volumes, software-based medical devices are experiencing exponential growth. The FDA's Center for Devices and Radiological Health (CDRH) maintains a dedicated registry of artificial intelligence and machine learning (AI/ML)-enabled medical devices, tracking the rapid integration of deep learning, computer vision, and predictive algorithms into clinical workflows.

According to the FDA's registry and independent tracking by Innolitics, the FDA authorized a record 295 AI/ML-enabled medical devices in calendar year 2025, representing a 16.6% increase over the 253 devices cleared in 2024, and a 33.5% surge over the 221 devices cleared in 2023. This brings the cumulative total of cleared AI/ML-enabled devices to approximately 1,451 as of early 2026.

This software surge is characterized by several distinct regulatory and technological trends:

  • Pathway Selection: The overwhelming majority—between 94% and 95% of all AI/ML-enabled devices—enter the market via the 510(k) pathway. Only a small fraction route through the De Novo pathway (for novel software without a predicate) or the PMA pathway (typically for high-risk software, such as diagnostic algorithms for critical cardiac conditions).
  • Dominance of Radiology: Radiology is the primary entry point for medical AI. About three-quarters of cleared AI/ML devices sit in the Radiology panel — Innolitics pegs the 2025 cohort at 71.5% and IntuitionLabs puts the cumulative list at roughly 76% — consisting of computer-aided detection (CADe) and computer-aided triage (CADt) software designed to analyze CT scans, MRIs, and mammograms for stroke, fractures, or oncology indicators.
  • Cardiovascular and Neurology Expansion: Beyond radiology, cardiovascular software (such as ECG analysis algorithms and echocardiogram image characterization) represents the second-largest panel, followed by neurology applications.
  • SaMD vs. SiMD: Approximately 62% of these clearances represent Software as a Medical Device (SaMD)—standalone software running on commercial off-the-shelf hardware or cloud servers. The remaining 38% represent Software in a Medical Device (SiMD)—embedded software that controls physical hardware, such as adaptive algorithms inside ultrasound machines or ventilators.
  • Review Timelines: According to Innolitics' analysis of 2025 AI/ML clearances, the median time from FDA submission to decision was 142 days. This represents a highly competitive review window, although it varies significantly based on whether clinical performance data or human factors validation is required.
  • Applicant Fragmentation: The AI/ML landscape is highly fragmented compared to traditional medtech. The 295 clearances in 2025 were distributed across 221 unique manufacturers, demonstrating that small software developers and digital health startups can successfully clear products without the capital scale required for physical hardware.

The rapid growth of AI/ML clearances presents unique challenges for payers and providers. Because the 510(k) pathway focuses on establishing "substantial equivalence" to a predicate, cleared software is not required to demonstrate superior clinical outcomes or long-term therapeutic utility. This creates a significant "evidence gap" between regulatory clearance and commercial reimbursement. Payers are increasingly skeptical of software clearances that lack robust clinical trial data, often refusing to issue dedicated coverage policies or reimbursement codes. For a detailed breakdown of how digital health developers are navigating this clearance-to-payment divide, read our analysis on digital health endpoints: FDA DHT qualification and the payer gap.


What is happening with De Novo, Breakthrough Devices, and the eSTAR mandate in 2025–2026?

While the 510(k) and PMA pathways serve established technology categories, novel medical devices that have no legally marketed predicate must seek authorization through alternative routes. The primary gateway for novel, low-to-moderate-risk devices is the De Novo classification pathway. If a device is automatically classified as Class III because it has no predicate, but the clinical risk profile does not justify the intensive demands of a PMA, the manufacturer can file a De Novo request to down-classify the technology to Class I or Class II.

However, recent trade reporting from Fierce Biotech indicates that the De Novo pathway has experienced a significant slowdown in 2025 and 2026. The FDA granted only 3 De Novo requests in the first quarter of 2025, with 46 applications sitting in active review or backlog. This slowdown is driven by several converging factors:

  1. The eSTAR Submission Mandate: Effective October 1, 2025, the FDA mandated that all De Novo classification requests must be submitted using the electronic Submission Template and Resource (eSTAR). eSTAR is an interactive PDF template that guides applicants through the exact data fields, standards, and attachments required for a complete submission. While eSTAR is designed to eliminate administrative "refuse to accept" (RTA) decisions and standardize reviews, the transition has created a steep learning curve for smaller novel-device developers, delaying initial filings.
  2. CDRH Resource Constraints: The implementation of the Medical Device User Fee Amendments (MDUFA V) has shifted CDRH priorities toward meeting performance goals for standard 510(k)s and PMA supplements, leaving the highly complex, resource-intensive De Novo reviews with extended queue times.
  3. Premarket Review Rigor: The FDA has increased its expectations for clinical validation data within De Novo submissions, frequently requiring prospective, multi-center trials that resemble Class III PMA requirements, rather than relying solely on bench testing or literature reviews.

The Breakthrough Devices Program: A Pipeline of Priority Reviews

To offset these development bottlenecks for technologies that address unmet clinical needs, the FDA operates the Breakthrough Devices Program. This voluntary program provides manufacturers with intensive interactive feedback, priority review, and senior management involvement during the premarket development phase.

As of March 31, 2026, the FDA's Center for Devices and Radiological Health (CDRH) and Center for Biologics Evaluation and Research (CBER) have granted 1,284 Breakthrough Device designations, including devices originally designated under the Expedited Access Pathway (EAP). CDRH granted 1,264 of those and CBER granted 20. Of the designated devices, 198 had received marketing authorization as of the same date (193 via CDRH and 5 via CBER) — meaning only about 15% of designated devices have crossed into the market, underscoring how early in development most designations are granted.

FDA reports the designations by clinical advisory panel, and three panels account for the bulk of the pipeline:

Breakthrough Device Designations by Clinical Panel (as of March 31, 2026)

Clinical Panel Approximate Designations Representative Target Technologies
Cardiovascular ~260 Next-generation stents, transcatheter valves, heart-failure monitoring, circulatory support.
Neurology ~200 Brain-computer interfaces, neuromodulation for paralysis and stroke recovery, epilepsy devices.
Orthopedic ~176 Joint-replacement implants, spinal fixation, trauma plates, regenerative bone scaffolds.
Gastroenterology & Urology ~100+ Endoscopic devices, continence therapies, GI diagnostic instruments.
All other panels (combined) remainder In vitro diagnostics, general/plastic surgery, microbiology, ophthalmology, digital health.

Data Source: FDA Breakthrough Devices Program page and program metrics (March 31, 2026 update). Panel figures are FDA-reported specialty counts; lower-ranked panels are summarized because FDA publishes the full breakdown only on the program page.

The Breakthrough program has succeeded in accelerating development timelines, but it does not alter the FDA's statutory standards for market authorization. A Breakthrough device must still submit a full 510(k), PMA, or De Novo application and demonstrate safety and effectiveness.

Furthermore, the commercial value of a Breakthrough designation was historically constrained by the lack of automatic reimbursement. However, the proposed Transitional Coverage for Emerging Technologies (TCET) pathway aims to establish immediate, temporary Medicare coverage for FDA-designated Breakthrough devices upon market authorization, creating a powerful commercial incentive for manufacturers to secure designation early in the clinical trial phase.


Frequently Asked Questions

How many 510(k) clearances does FDA grant per year?

The FDA clears between 3,000 and 3,300 devices under the 510(k) pathway each calendar year. The database recorded 3,225 clearances in 2025, 3,129 in 2024, and 3,346 in 2023. These numbers align closely with the FDA's official fiscal-year figures (3,238 in FY2025 and 3,107 in FY2024).

What share of FDA device authorizations are PMA versus 510(k) versus De Novo?

The vast majority of premarket authorizations route through the 510(k) pathway, representing approximately 98.5% of combined annual clearances and approvals. Original PMA approvals for high-risk Class III devices make up only about 1% to 1.5% of the total (typically 30 to 45 approvals per year). The De Novo pathway represents a small fraction, usually accounting for fewer than 30 to 50 grants annually.

How many AI/ML-enabled medical devices has FDA cleared?

As of early 2026, the FDA has authorized approximately 1,451 AI/ML-enabled medical devices. The volume reached a record 295 authorizations in calendar year 2025, up from 253 in 2024. Radiology is the dominant clinical panel, representing about three-quarters of AI/ML clearances (71.5% of the 2025 cohort per Innolitics; roughly 76% of the cumulative list per IntuitionLabs), and SaMD (Software as a Medical Device) makes up roughly 62% of the cohort.

What is the difference between a PMA original and a PMA supplement?

An original PMA is the first-time premarket approval application filed for a new Class III medical device, requiring extensive clinical trial data to demonstrate safety and effectiveness. A PMA supplement is a subsequent application filed to modify an already-approved device. Because Class III change-control is highly regulated, manufacturing, design, labeling, or location changes generate a large volume of supplements, which make up 97.4% of all PMA records in the FDA database.

Has the De Novo pathway slowed down in 2025–2026?

Yes. The De Novo pathway experienced a significant slowdown in 2025, logging only 3 grants in the first quarter of the year with dozens of submissions backlogged. This contraction is attributed to the implementation of the mandatory eSTAR electronic submission template on October 1, 2025, increased FDA demands for clinical trial data, and resource constraints at CDRH under the current MDUFA V agreement.


Sources

  1. U.S. Food and Drug Administration (FDA). 510(k) Clearances Database and Program Page. FDA 510(k) Clearances
  2. U.S. Food and Drug Administration (FDA). Premarket Approval (PMA) Database and Program Page. FDA Premarket Approvals
  3. U.S. Food and Drug Administration (FDA). Artificial Intelligence and Machine Learning (AI/ML)-Enabled Medical Devices. FDA AI/ML Device List
  4. U.S. Food and Drug Administration (FDA). Breakthrough Devices Program and Public Dashboard Status. FDA Breakthrough Program
  5. openFDA Medical Device Databases API. 510(k), PMA, and Classification Endpoint Exports (July 6, 2026). openFDA Medical Device API
  6. ASME Journal of Engineering and Medical Systems. A Decade of Food and Drug Administration Medical Device Authorizations (2015–Jun 2025). ASME Digital Collection
  7. Innolitics. 2025 Year in Review: AI/ML Medical Device 510(k) Clearances. Innolitics 2025 Analysis
  8. Fierce Biotech. FDA's novel medical device clearances slow to a crawl (De Novo pathway delays). Fierce Biotech MedTech News
  9. U.S. Code of Federal Regulations. 21 CFR Part 807 Subpart E — Premarket Notification Procedures (510k). 21 CFR Part 807
  10. U.S. Code of Federal Regulations. 21 CFR Part 814 — Premarket Approval of Medical Devices (PMA). 21 CFR Part 814
Ran Chen
Contributing Editor
Ran Chen

Founder, PharmaDossier. Life-sciences operator covering market access, specialty pharma, biosimilars, and regulated healthcare growth.

Follow on LinkedIn →