2026-08-09T18:30:00.000Z
Digital Pathology’s Growth Is Shifting From Scanners to Software
The 8.00% CAGR compounds on a base already tilted toward software, where AI-enabled platforms, not scanner hardware, are absorbing most of the new investment.
Fewer than 15% of U.S. health systems have chosen a digital pathology vendor, and the scanner is no longer the decision that matters most. Kaiso Research's primary dataset puts the global digital pathology market at $1.11 billion in 2024, climbing to $2.59 billion by 2035 at an 8.00% compound annual growth rate, and the software layer sitting on top of the scanner, not the glass and optics underneath it, is absorbing most of that growth. Roche, Philips, and Leica Biosystems built their franchises on hardware. PathAI and Proscia are building theirs on algorithms that run on somebody else's scanner.
That's not a subtle distinction for a hospital system, a pharma sponsor, or an investor deciding where the next dollar of pathology infrastructure spend actually goes. The pathologist workforce is not growing fast enough to keep pace with rising cancer caseloads, and every health system in this dataset is running the same math: digitize now, or fall further behind on turnaround time. Radiology went through this exact transition with PACS two decades ago. Pathology is going through it now, with a much smaller base of installed hardware standing in the way.
The $1.11 Billion Base Is Already Weighted Toward Software, Not Scanners
Software already captures more of the 2024 base than the scanners, slide management systems, and other hardware that most buyers still associate with going digital. This coverage segments the market by product into software and device, by application into academic research, disease diagnosis, and drug discovery and development, and by end use into hospitals, biotech and pharma companies, diagnostic labs, and academic and research institutes.
Disease diagnosis, concentrated in cancer cell detection, is the largest application segment, and it is the one forcing the software question fastest. A hospital lab running oncology, hematology, and infectious disease workups needs second opinions turned around in hours, not days, and a scanner sitting next to a microscope cannot do that on its own. Academic research is the second growth vector, driven by digital slide archives, biomarker studies, and virtual teaching tools that universities and pharmaceutical labs are building out together.
The report's scope also splits the market by pathology type, between human pathology and veterinary pathology, a distinction that matters more than the headline CAGR suggests. Human diagnostics carries the FDA and CMS machinery described later in this analysis. Veterinary labs do not, and that gap in regulatory scaffolding shapes which buyers can move fast and which cannot.
North America is the largest region in this coverage, Europe is second, and Asia Pacific is set to grow fastest through 2035 as China, India, Japan, and South Korea fund public hospital modernization programs. The pattern that emerges from Kaiso Research's primary market data is a market where hardware opened the door in the 2010s and software is doing most of the buying from here through 2035. That is the split every later section of this analysis returns to.
Roche, Philips, and Leica Own the Scanners. PathAI and Proscia Are Winning the Software Fight.
Ten vendors dominate this coverage of the market, and they split cleanly into two camps: the hardware incumbents that built the category, and the AI-native software companies now competing for the same hospital budget line. Leica Biosystems, owned by Danaher, F. Hoffmann-La Roche through its Ventana business, Koninklijke Philips N.V., Hamamatsu Photonics K.K., and 3DHISTECH built their positions on scanners and slide digitizers. Huron Digital Pathology and Inspirata occupy an adjacent tier, supplying image management infrastructure to labs that are not ready to commit to a single platform vendor.
Philips upgraded its IntelliSite Pathology Solution with AI capabilities in June 2024 to speed cancer diagnosis, and Roche followed with an AI-ready, cloud-compatible release of its uPath enterprise software in October 2023. Hamamatsu shipped the NanoZoomer S360 scanner in July 2023, built for high-volume labs that need speed and image quality above all else. Every one of these moves points the same direction: hardware vendors bolting AI and cloud connectivity onto platforms that used to compete purely on scan speed and image resolution.
PathAI, Proscia, and Indica Labs are approaching the same buyer from the other side, selling AI and workflow software that runs on scanners built by someone else. Indica Labs has carved out a research and pharma-facing niche in quantitative image analysis, competing for the drug discovery and development segment of the application breakdown rather than the hospital diagnosis segment PathAI and Proscia are targeting first. KLAS Research's 2026 report on U.S. digital pathology adoption found that pathology-specific vendors like Proscia, PathAI, and Lumea are currently outperforming traditional radiology vendors on early customer satisfaction, while Leica Biosystems remains the undisputed leader for clinical scanners in the U.S. The same report found that Roche's scanner customers report meaningfully lower satisfaction, citing reliability and throughput concerns, a gap that opens room for a software layer that does not care which scanner produced the image.
That gap won't stay open indefinitely. Buyers surveyed by KLAS described integration as non-negotiable: AI has to sit inside the image management workflow pathologists already use, not in a separate login they have to remember. The vendor that owns that workflow relationship, not necessarily the vendor that owns the scanner, is the one holding the renewal conversation five years from now.
Six FDA-Cleared Scanners Later, the Real Fight Moved to the Algorithm Layer
Six whole slide imaging systems now hold FDA clearance for primary diagnosis, and every one of them is fundamentally a scanner. A 2025 comparative review identified six FDA-cleared surgical pathology WSI scanners on the market: Epredia's E1000 Dx, Philips' IntelliSite Pathology Solution, Hamamatsu's NanoZoomer S360MD, two separate Leica Aperio systems, and Roche's VENTANA DP 200. Scan quality, rescan rates, and slide-handling reliability still vary meaningfully between them, but the clearance itself has stopped being the differentiator it was in 2017.
The differentiation buyers actually pay for now sits one layer above the glass, in the image management system and the AI models reading what the scanner captured. Cloud-native platforms are adding foundation-model layers, of which Proscia's Concentriq Embeddings is one example, that let algorithm developers build on pre-trained image representations instead of training every diagnostic model from a blank slate. That shortens the distance between a research finding and an FDA submission, and it is why the newest clearances are arriving faster than the scanner-era clearances did.
Interoperability with laboratory information systems and hospital electronic health records has become the practical gate on top of the regulatory one. A platform that cannot plug into the LIS a hospital already runs will lose the deal regardless of how strong its underlying algorithm is, because pathologists won't tolerate a second system for daily case review.
The next technical shift is multimodal. Vendors are pairing histology images with genomic, proteomic, and structured clinical data inside the same model, moving the platform from a slide viewer into a diagnostic reasoning layer. A pathologist who once needed a separate genomics report and a separate slide review is closer to getting both inside one interface, which raises the stakes on choosing a platform built to expand rather than one built only to scan.
One Predicate Device From 2017 Still Shapes Every Later Approval
The FDA's first whole slide imaging clearance for primary diagnosis went to a single company in 2017, and that decision still constrains every later entrant. On April 12, 2017, the FDA granted Philips a De Novo classification for its IntelliSite Pathology Solution, the first and at the time the only WSI system cleared in the U.S. for primary diagnostic use. That clearance downgraded whole slide imaging from a Class III to a Class II device category and made the Philips system the predicate for every 510(k) submission that followed it. Outside the U.S., the CE-IVD mark had already permitted primary diagnosis on WSI systems since 2014, three years ahead of the American market.
This coverage points to the same regulatory logic driving adoption today that drove it after 2017: FDA clearance lowers the risk of a lab's first digital deployment, and CMS reimbursement makes that deployment financially defensible once it is running. European buyers are working through a comparable process, with EU-level funding supporting teleconsultation infrastructure and a slow harmonization of data privacy rules across member states that determines how easily a scan taken in one country can be read in another.
None of this required a new law. It required one company willing to run the De Novo process first, and a regulator willing to build a predicate pathway around what that company proved. Hamamatsu's scanner, both Leica Aperio systems, Roche's VENTANA DP 200, and PathAI's software all cleared the same pathway that single 2017 decision opened.
Caseloads, Workforce, and Reimbursement Are Now Pulling in the Same Direction
]Three forces are converging on the same buying decision, and none of them would move the market on its own. Cancer incidence keeps climbing across every region in Kaiso Research's coverage, which by itself would simply mean more slides for a shrinking pool of pathologists to read. That shortage is the real constraint. It is what turns a nice-to-have imaging upgrade into a workforce necessity, because a lab that cannot hire its way out of a backlog has to digitize its way out instead.
Reimbursement removes the second constraint, and it removes it unevenly. In the U.S., the same FDA clearance pathway described above sits alongside CMS reimbursement decisions that determine whether a digitized workflow pays for itself inside a hospital's existing billing structure. In Europe, reimbursement varies by country even as CE-IVD marking stays constant across the bloc, so a platform's commercial rollout in Germany does not automatically translate into revenue in Spain or Italy.
Regulatory trust in the algorithms is the third force, and it is the newest one. A lab willing to digitize slides in 2020 was betting on image quality and storage economics. A lab evaluating a platform in 2026 is betting on an AI model's diagnostic accuracy, and that bet only gets easier to make as more of PathAI's, Ibex's, and Paige's tools clear FDA review.
Each of these three forces would slow the market meaningfully if it moved alone. Together, they are why the forecast holds an 8.00% compound rate through 2035 rather than something closer to the low single digits typical of a mature medical device category.
Three Regulatory Firsts in Eleven Months Show Where the Frontier Is Moving
PathAI cleared three separate regulatory milestones between June 2025 and March 2026, and together they map where digital pathology approvals are headed next. PathAI received 510(k) clearance for its AISight Dx platform for primary diagnosis in June 2025, a cloud-native image management system cleared for use with the Hamamatsu NanoZoomer S360MD and Leica Aperio GT 450 DX scanners and CE-IVD marked across the EEA, UK, and Switzerland. That clearance mattered because it proved a software-first vendor, not a scanner manufacturer, could clear the same primary-diagnosis bar Philips set in 2017.
Eight months later, in March 2026, the FDA granted PathAI Breakthrough Device Designation for PathAssist Derm, an AI tool built to analyze whole slide images of skin lesions and support dermatopathology review. In the same period, PathAI's AIM-MASH AI Assist earned joint EMA and FDA qualification as the first AI-powered pathology Drug Development Tool, a distinct regulatory category built for algorithms used inside pharmaceutical trials rather than clinical diagnosis.
Three clearances in three different regulatory categories inside eleven months is not a pattern any single hardware vendor in this market has matched. It is a signal that regulators have built enough precedent around AI-assisted pathology to move faster on the next submission than they did on the first one.
The precedent goes back further than PathAI's 2025 clearance. Paige Prostate became the first FDA-cleared AI tool for pathology in 2021, and Ibex Medical Analytics followed with clearance for its own prostate cancer detection algorithm, Ibex Prostate Detect. Both cleared before most hospital buyers had settled on a scanner, let alone an AI vendor, which is exactly why the eleven-month stretch above reads as acceleration rather than a first step.
Capital Is Following the Software Layer, Not the Hardware
Two transactions in 2024 and 2025 show where money is actually flowing inside this market, and neither of them went toward a scanner company. Proscia raised $50 million in March 2025, in a round led by the software investor Insight Partners alongside AI Capital Partners, Triangle Peak Partners, and several smaller funds. The raise brought Proscia's total funding to $130 million and followed a year in which the company says its Concentriq platform grew to serve 16 of the top 20 pharmaceutical companies, on track to support more than 22,000 patient diagnoses a day.
Quest Diagnostics took a different route into the same trend. Materials from Quest's 2025 Investor Day disclosed eight acquisitions completed in 2024 that were expected to generate roughly $1 billion in combined annualized revenue, and one of them was PathAI Diagnostics' Tennessee laboratory, a digital pathology asset folded directly into Quest's diagnostic lab network. A national reference lab operator did not just license AI software. It bought the lab infrastructure that runs on it.
Those are two different bets on the same conclusion: whoever controls the software and the workflow around a pathology slide captures more of the value than whoever manufactured the scanner that produced the image.
What Health Systems and Platform Investors Should Model Before They Sign
Health system buyers and platform investors are reading this market for different signals, and both groups are underpricing the same risk. A hospital system evaluating its first digital pathology platform should model the total cost of the combined scanner, image management system, and AI stack, not the scanner's throughput in isolation, because the vendor chosen first constrains which AI models the lab can add later. Acting inside the current window matters. With fewer than 15% of U.S. health systems having selected a vendor, an early buyer still has real influence over a platform's roadmap instead of adopting whatever a later majority settles on.
Investors and platform vendors face the mirror version of the same problem. Capital is concentrating in software companies that already hold FDA clearance and can point to health system references, not in point-solution AI vendors waiting on their first approval. This market will consolidate around a handful of platforms that combine regulatory clearance, LIS integration, and pharma-scale distribution. Vendors without at least one of those three advantages are running out of runway to build it before a better-capitalized competitor closes the gap.
The decision quality gap between an informed buyer and an uninformed one is widening, not narrowing, as the vendor landscape gets more crowded.
The Closed-System Rule Is the Risk Few Buyers Are Pricing In
Three structural risks sit underneath the growth numbers, and the first one is regulatory. An FDA-cleared whole slide imaging system is a closed unit: the scanner, the viewing software, and the display monitor that received clearance together cannot be swapped for unapproved components without triggering a new submission. A hospital that clears a Philips scanner paired with one AI vendor's software cannot simply plug in a competitor's algorithm later without both companies revisiting their regulatory paperwork. That is a real switching cost, and it rarely appears in a vendor's sales pitch.
The second risk is fragmentation. This coverage lists ten meaningful vendors competing inside a market this dataset puts at $1.11 billion in 2024, and KLAS's finding that fewer than 15% of U.S. health systems have committed to a vendor means most of that base is still contestable. Some of today's smaller AI-native vendors will not survive to the other side of a consolidation wave, which turns an early platform choice into a bet on that vendor's balance sheet as much as its algorithm.
The third risk sits inside the report's own segmentation and rarely gets acknowledged: the report's scope spans both human and veterinary pathology, and the two adopt on entirely different timelines. Veterinary labs face none of the FDA primary-diagnosis pathway described above, which means the human-diagnostics growth story driving most of this forecast does not automatically extend to the veterinary segment sitting inside the same market definition.
The Decision Window Is Open, and Most Buyers Have Not Walked Through It
Every thread in this analysis points to the same conclusion: value in digital pathology has moved from the device that captures the image to the software that interprets it, and the market has not finished repricing that shift. Roche, Philips, and Leica built durable businesses on scanners that will keep selling for years. PathAI, Proscia, and a small group of AI-native challengers are building the layer that decides which of those scanners gets chosen next.
Regulatory clearances are compounding faster than they did a decade ago, capital is following the software layer rather than the hardware, and fewer than one in seven U.S. health systems has made a vendor decision that will be expensive to reverse. The forecast window tracked here through 2035 will reward whoever recognizes that timing now, not whoever waits for the market to look obvious.
Watch the next KLAS survey cycle for the real signal. If vendor selection climbs meaningfully past 15% before 2028, this market has already sorted itself into winners. If it has not, the software layer described throughout this analysis is still up for grabs.
The scanner stopped deciding. Most have not caught up.
