UmbriaExplore our travel guides. Share your stories, tips and questions on Reddit.
Fast movers • Alerts • Momentum
10% OFF with SAVE10
Charts • Watchlists • Multi-broker tools
Built for active traders
Biotech catalyst, news and analysis PDUFA tracker
Biotech catalyst, news and analysis PDUFA tracker
FDA De Novo authorization opens the U.S. burn-care market. Hospital adoption, clinical outcomes, payment and cash consumption now determine the next stage.
Get every Merlintrader report in real time on Telegram: join @merlintraderpub_com.
MDAI common stock; MDAIW is a separately traded warrant. No product revenue reported in the Q2 filing.[2]
The August results release targets first U.S. sales by year-end 2026 and initiation of the Triage and Treatment Outcome Study in Q4. These are company objectives, not a fixed FDA decision date. The October label-expansion study is expected to complete in 2027.[5][7]
At June 30, cash and equivalents were $13.980 million, notes payable carrying value was $14.914 million and H1 operating cash use was $7.825 million. Management’s one-year liquidity assessment combines cash, BARDA commitments and Yorkville availability; it is not a cash-only guarantee. DeepView remains an adjunct to clinical assessment.[2][3]
Spectral AI’s question has moved beyond whether the initial burn-care device can obtain U.S. authorization. The May 21 De Novo order is granted; pricing, paid uptake and recurring economics must now develop. The current financial base remains government-funded research, with $86.6 million committed under a contract capped at approximately $150 million and a $9.7 million company development contribution. A head, hands and feet study adds a future evidence pathway, not an expanded label already granted.[2][3][7]
The October 5–8 Chicago assembly is now past. Spectral AI confirmed product display and engagement with emergency-department decision makers, without announcing an order.[8]
Company expects an August 2027 CMS determination; if approved, payment status would generally start in October 2027. Submission is not payment approval.[6]
First site initiation supports research on head, hands and feet burns. Blinded observational training and validation study expected to complete in 2027.[7]
Company announced a presentation led by James H. Holmes IV for the September 26–29 congress. The scheduled event is past; this announcement is not a subsequent presentation-results report.[10]
Editorial assessment of financial and operational robustness over the next twelve to eighteen months, dated October 13, 2026. Each pillar is scored from 1 (fragile) to 5 (robust).
| Pillar and weight | Score / 5 | Basis |
|---|---|---|
| Financial resources · 30% | 2.5 | Cash of $13.980 million; carrying debt of $14.914 million; resource-based liquidity assessment depends partly on contract commitments and equity access.[2] |
| Catalysts · 30% | 3.0 | First sales and outcomes work are company targets. Anatomy expansion and NTAP have future clinical or CMS dependencies.[5][6][7] |
| Capital allocation · 20% | 2.0 | A $9.7 million development contribution, commercial spending, financing warrants and equity capacity compete for a commercial return.[2] |
| Trading liquidity · 10% | 2.0 | October 13 common-stock volume was 178,183 shares on a roughly $50.21 million vendor capitalization. Liquidity can vary materially.[11] |
| Operating execution · 10% | 2.5 | FDA authorization is obtained, but product revenue was absent in the June filing and the financial-control material weakness persisted.[2][3] |
Weighted calculation: (2.5 × 30%)+(3.0 × 30%)+(2.0 × 20%)+(2.0 × 10%)+(2.5 × 10%) = 2.50. This is not an indication to buy or sell, a target price or a judgment about the share price. It is an editorial description of robustness, not an investment recommendation.
The full deep dive has the answer’s building blocks: cash, dilution, catalysts and risks, every figure sourced.
Free. No signup. You decide, we don’t recommend.
Spectral AI has crossed the most important regulatory threshold for its initial U.S. burn-care application. The FDA’s May 21, 2026 De Novo order permits marketing of the DeepView AI System for its specified adjunctive thermal-burn assessment use. Management’s next commercial objective is first U.S. sales by year-end 2026. A constructive operating scenario would combine those first sales with installations, staff training, routine utilization and renewed software relationships. The commercial value would come from repeated use by customers, rather than from the regulatory announcement alone.[3][5]
The government-supported deployment path could help bridge the gap between a clinically developed product and hospital adoption. The September investor presentation describes subsidized procurement for an initial 30 U.S. systems and planned deployment of up to 30 systems through June 2027. These are company plans and contractual program descriptions, not evidence that all 30 systems have already been sold or installed. Their economic significance would become clearer as delivery, acceptance, revenue recognition and clinical utilization emerge. Successful early placements could also provide the experience needed to support hospitals considering purchases outside the subsidized program.[4]
In this scenario, the device-and-software model would gradually change the revenue mix. A hardware placement can create an initial transaction, while a software relationship can potentially continue after that placement. The company’s presentation describes recurring software and services and a three-year minimum model, but also says contract structure, term and pricing remain subject to negotiation and that pricing has not been finalized. The constructive outcome therefore depends on negotiated economics and retention, not merely on multiplying a planned installed base by an assumed annual fee.[4]
A middle scenario would include initial commercial activity without an immediate transformation of the financial statements. Research and development revenue could remain the dominant revenue category while hospitals take time to evaluate budgets, clinical workflows, IT requirements and purchasing processes. Spectral AI itself describes different adoption routes: existing clinical sites can move faster, while committee and AI/IT review can take six to twelve months in the illustrated commercial model. Those timelines explain why clinical interest and sales-team activity may precede recognized revenue by a meaningful interval.[4]
Government development work and commercial preparation would coexist in this scenario. The second-quarter release attributes declining reimbursed research revenue and gross margin partly to the narrowing scope of work after FDA authorization and the follow-on program’s company cost share. Meanwhile, sales and marketing expense increased ahead of first commercial sales. A transition can therefore produce an awkward combination: the original development program becomes less revenue-generative before the commercial organization has created a large replacement stream. The key issue is whether that interval remains financially manageable.[5]
The cautious operating scenario is not a reversal of the May authorization. It is a slower conversion of authorization into customers and cash. Hospitals could find the tool clinically useful but take longer to fund it, reimbursement could be less supportive than management hopes, or rollout costs could exceed early product revenue. The annual report identifies pricing, reimbursement, manufacturing, adoption and financing as material risks. An NTAP application offers a possible future payment route, but it is not an approved payment entitlement.[1][6]
In that scenario, the company’s own development contribution and commercial spending would compete for cash while debt service continues. The June balance sheet carries $14.914 million of notes payable, against $13.980 million of cash and equivalents, and the company used $7.825 million in operating cash during the first half. The liquidity assessment also relies on contract commitments and equity-facility availability. Slower sales would therefore matter both operationally and to the number of shares through which future business value is divided.[2]
Spectral AI, Inc. is a Delaware corporation headquartered in Dallas, Texas. Its common stock trades on Nasdaq under MDAI; separately listed redeemable warrants trade under MDAIW. The second-quarter Form 10-Q identifies both classes and their different instruments. MDAI is an equity interest in the issuer. MDAIW is a contractual right governed by warrant terms, including its exercise price and expiration provisions. The warrant’s price, trading volume and potential leverage are not the common stock’s market data.[2]
The operating business is predictive medical diagnostics built around proprietary multispectral imaging and artificial intelligence. The current commercial focus is burn assessment. The consolidated group includes Spectral MD Holdings LLC, Spectral MD, Inc., Spectral MD UK Limited and Spectral IP, Inc. The October company releases explicitly clarify that the BARDA contract is held and performed through wholly owned Spectral MD. Reporting at the listed parent consolidates these activities; the government contracting party and the publicly traded legal entity remain distinct.[2][8]
The company’s financial reporting uses U.S. GAAP and U.S. dollars. Quarterly figures are unaudited interim statements; the 2025 annual financial statements were audited by Forvis Mazars. The annual auditor’s opinion addresses the consolidated financial statements, not an audit opinion on the effectiveness of internal control. That distinction matters because the company separately reports a material weakness in its financial reporting controls. A financial statement audit and a conclusion that controls are effective are different assertions.[1][2]
The public-company history also matters to the capital structure. The 10-Q describes the September 2023 business combination with Rosecliff Acquisition Corp I and the public warrants assumed in that transaction. Those legacy warrants remain part of the potential share supply, alongside subsequent financing warrants, options and restricted awards. An assessment of the company’s operating progress needs to sit beside an assessment of this financing history: development funding, debt proceeds and equity issuance have all contributed to the capital available for commercialization.[2]
Spectral AI is not simply a generalized AI-software business. Its product depends on a proprietary imaging device, clinical image-and-outcome data, a medical algorithm, regulatory authorization and adoption by healthcare providers. Each component creates a different execution requirement. Hardware must be manufactured and supported; the algorithm must perform within its validated use; hospitals must incorporate the output into clinical decisions; and the commercial model must capture sufficient economics from those activities. The company’s filings describe these interlocking dependencies rather than a frictionless software rollout.[1][2]
The DeepView System combines multispectral imaging with proprietary AI to assess healing potential within burn wounds. The imaging process captures information beyond ordinary visual inspection and produces an overlay highlighting areas predicted not to heal with conservative treatment alone within 21 days. Company descriptions give an image-acquisition time of 0.2 seconds and approximately 20 to 25 seconds for image processing and AI classification. Those are product-process descriptions; a complete clinical encounter includes positioning, preparation, interpretation and clinical decision-making.[1][8]
The FDA order provides the controlling description of the authorized output. DeepView AI-Burn software analyzes images to detect deep burn tissue, including deep partial-thickness and full-thickness tissue. It provides segmentation of areas unlikely to heal within 21 days with conservative treatment alone, and the system calculates wound area and percentage total body surface area burn. An absence of highlighted tissue indicates an absence of deep burn tissue under the described software output. The authorization is tied to this intended use, rather than to every possible wound-diagnostic application.[3]
The device is an aid to physicians managing thermal burn wounds and considering treatment plans. It must be used alongside clinical assessment, not as a standalone diagnostic device. The order also states that the segmentation contour does not give definitive delineation of the deep-burn margin for precise excision sizing or the exact borders of non-healing tissue. That is a practical boundary on the product’s role: predictive information can contribute to a treatment plan without becoming an autonomous instruction for where a surgeon should cut.[3]
The distinction between prediction and treatment is central. A diagnostic algorithm can improve information available to a clinician while leaving the ultimate intervention dependent on the patient’s condition, wound location, timing and the clinician’s broader assessment. Spectral AI’s clinical and commercial argument is that earlier objective information can improve decisions. Whether it consistently produces shorter hospital stays, fewer unnecessary procedures or better outcomes is a related but separate question from whether its classification output predicts healing accurately.[3][9]
The company reports a proprietary clinically validated database of more than 340 billion pixels of burn image data. Pixels describe the scale of information available to the model; they are not 340 billion patients or independent clinical observations. The value of such a repository depends on image quality, representative patients, reliable outcome labels and separation of training from validation. The October label-expansion study explicitly distinguishes training from independent validation, illustrating why the structure of the evidence matters more than the largest numerical description of the data bank.[7][8]
FDA granted De Novo classification through order DEN250028 dated May 21, 2026. The device is classified into Class II under 21 CFR 878.1840, with product code SHY and the generic category of software-aided adjunctive diagnostic devices for healthcare providers assessing skin wounds. The company’s public announcement followed on May 26.[12] The regulatory decision date and announcement date are therefore five days apart. Both are relevant, but they describe different events.[3]
The De Novo pathway establishes a regulatory classification for a novel device type with general and special controls. The FDA letter permits immediate marketing of the described device subject to those controls. This was a De Novo classification order, not a PMA approval and not an ordinary substantial-equivalence 510(k) clearance. Company communications often use the shorthand “FDA-cleared”; the underlying event is the granted De Novo request and its specific indications for use.[3][5]
The order identifies risks associated with incorrect results, use errors, malfunction, electrical or thermal exposure and interference with other devices. Required controls include clinical and non-clinical performance testing, precision testing, human factors, software verification and validation, safety testing and labeling. Clinical performance testing must demonstrate improvement in device-aided diagnostic characterization compared with unaided users, evaluate patients across representative risk factors and assess standalone output against ground truth. The controls are part of the authorization’s continuing framework, not merely paperwork completed before launch.[3]
For devices mapping wounds, the special controls require segmentation-accuracy testing. Labeling must summarize clinical and standalone performance, sensitivity and specificity, confidence intervals and clinically relevant subgroup performance. It must also identify limitations and subpopulations where performance may differ, explain interpretation of outputs and state that the product is not intended as a standalone diagnostic or to confirm a clinical diagnosis. These requirements place the device’s strengths and limits within the information clinicians receive.[3]
Authorization does not remove manufacturing, reporting and post-market obligations. FDA expressly notes that the granted request is not a determination of compliance with every other requirement of the Food, Drug and Cosmetic Act. Registration, listing, labeling, adverse-event reporting, quality-system requirements and unique device identification remain applicable. For the business, obtaining the order changes the commercialization opportunity; maintaining a compliant medical-device operation remains an ongoing cost and execution responsibility.[3]
The second-quarter 10-Q states that the validation study completed enrollment of 164 patients, including 49 pediatric subjects. It reports superiority in sensitivity and achievement of the non-inferiority margin in specificity compared with clinician assessment in that population. These are distinct statistical claims. Sensitivity addresses identification of non-healing regions; specificity addresses avoidance of incorrectly identifying healing tissue as non-healing. Improvement in one measure and a non-inferiority result in another do not amount to superiority on every diagnostic endpoint.[2]
The filing also describes three larger multisite U.S. clinical studies involving more than 400 adult and pediatric burn patients. The larger development history provides context, but the 164-patient validation cohort remains its own analysis. Cohort sizes, training images, validation images and accumulated image pixels are different denominators. Combining them into a single patient count or attaching a result from one cohort to another would blur the evidence rather than strengthen it.[2]
A separate prospective multicentre UK evaluation was published in Burns, DOI 10.1016/j.burns.2025.107550. Its original abstract describes 40 adults and 70 burn images at Newcastle upon Tyne and Manchester Adult Burn Centres, with admission imaging and follow-up through 21 days. Ground truth was clinical healing assessment at 21 days; ImageJ was used to compare pixels in AI-generated wound maps against healed and unhealed classifications. It was an observational product evaluation, not a randomized trial assigning treatment based on DeepView.[9]
That UK evaluation reported sensitivity of 80.7%, with a 95% confidence interval of 51.8% to 100%; specificity of 95.5%, with a confidence interval of 93.3% to 97.8%; and overall accuracy of 95.3%, with a confidence interval of 93.2% to 97.6%. These results describe that study’s images, patients and methods. The broad sensitivity interval is especially important when interpreting how precisely the estimate is known. A high overall accuracy can coexist with much greater uncertainty in the sensitivity estimate.[9]
The same abstract reports a mean scan-to-result time of five minutes and twelve seconds. That is not the same measure as the company’s 0.2-second acquisition and 20-to-25-second processing descriptions for the current system. They arise from different contexts and definitions of time. Neither number needs to displace the other; a technology’s internal processing speed and its observed clinical workflow time can be materially different.[8][9]
The UK paper’s authors conclude that the evidence base remains limited and requires further development, specifically including effects on clinical decisions, workflow integration and patient outcomes. That conclusion supports a measured interpretation: diagnostic performance can be encouraging while outcome evidence is still developing. It also explains the business relevance of the planned Triage and Treatment Outcome Study, which management describes as intended to assess treatment decisions, surgical precision and the care journey rather than simply repeat a headline accuracy figure.[5][9]
On October 1, 2026, Spectral AI announced completion of the first site initiation visit for a study intended to support expansion of DeepView’s burn indication to the head, hands and feet. The visit took place at Pinderfields Hospital in Wakefield, UK, led by principal investigator Dr. Preetha Muthayya. The company said completion of applicable site-readiness requirements enabled the hospital to begin enrollment. Site initiation is an operational launch milestone; it is not a completed study or a granted label expansion.[7]
The study is prospective training and validation research involving adult and pediatric patients with qualifying thermal burns in those anatomical areas. Imaging is planned within 72 hours of injury, with healing outcomes assessed through 21 days. The resulting data are intended to support algorithm training and independent validation. DeepView is to operate in a blinded observational capacity, not to direct patient care. The company expects multiple UK sites and study completion in 2027.[7]
These design choices explain both the opportunity and its limit. Hands, feet and head or face burns raise specific assessment challenges, so additional evidence in those regions could broaden the platform’s clinical utility. However, data collection, algorithm development, independent validation and a regulatory decision are separate steps. The October announcement describes the work meant to support a potential future expansion; it does not report authorized expanded use in the United States or completed clinical performance results from that new study.[7]
A separate Triage and Treatment Outcome Study was targeted for initiation in the fourth quarter of 2026 in the August results release. Management described its purpose as demonstrating improved surgical precision, faster treatment decisions, a better care journey and reduced length of stay. Those are intended study questions and company objectives. They are not outcomes already demonstrated by the October head, hands and feet site-initiation announcement, which describes a different blinded observational research program.[5][7]
The timing of the two projects can affect the commercial case differently. An anatomy-focused label study addresses the scope of use and performance in an expanded population. An outcomes study addresses the practical consequences of using information in treatment workflows. Both can be useful for adoption, but neither substitutes for hospital budgets, negotiated prices or payment policies. The strongest future evidence would connect diagnostic utility, clinical implementation and economics without collapsing those three layers into a single claim.[5][7]
The Project BioShield contract executed in September 2023 has a maximum value of approximately $150.0 million. The original award was approximately $54.9 million, with additional options for follow-on development and procurement. In March 2026, BARDA exercised part of those options, accelerating approximately $31.7 million and extending the base phase without additional cost through June 2026. The second-quarter filing reports $86.6 million committed to date, leaving approximately $63.4 million of further options at the government’s discretion.[2]
The arithmetic is straightforward: $54.9 million plus $31.7 million equals $86.6 million; the difference between the $150.0 million ceiling and that committed amount is $63.4 million. The interpretation is more important. The ceiling is the maximum potential contractual scope if relevant options are exercised. The committed amount describes funding obligations under the program. Neither is the company’s unrestricted bank balance, and neither can simply be booked as current-period revenue. Revenue follows performance and the applicable accounting policy.[2]
The contract is held and performed by Spectral MD, a wholly owned subsidiary. October company releases explicitly apply references to contract awards, revenue, opportunities and obligations to that subsidiary unless otherwise specified. Consolidation brings the activity into Spectral AI’s financial statements, but the legal party undertaking the work remains Spectral MD. This matters when assessing who bears contractual duties and when describing the government relationship.[8]
The follow-on funding includes a company commitment to provide $9.7 million of overall development costs associated with the feature advancements. It is therefore non-dilutive government support accompanied by a meaningful company contribution, not a program under which every dollar of development expenditure is reimbursed. The August results release links the cost-share phase to lower gross margin, as some costs continue while the proportion reimbursed under the program falls.[2][5]
First-half 2026 research and development revenue from BARDA was $6.589 million, compared with $11.030 million in the first half of 2025. Other U.S. governmental authorities contributed $0.926 million, bringing total first-half revenue to $7.515 million. BARDA’s contribution was approximately 87.7% of the total, calculated from those reported amounts. This customer concentration is substantial even though government development funding has reduced reliance on equity for parts of the technology program.[2]
Federal commitments also carry execution and policy risk. The annual report discusses dependence on government contracts, funding, performance requirements and the possibility of termination or changing priorities. A program can provide a credible development pathway while still leaving a company exposed to timing and budget decisions beyond its control. Commercial diversification would reduce that concentration only when customers outside the development program generate actual, repeatable revenue.[1][2]
Spectral AI had generated no product revenue as of the second-quarter filing. Its revenue then came from contract development and research services, principally U.S. government programs. Management targeted first-ever U.S. commercial sales by year-end 2026. The difference between those statements defines the stage of the business: an authorized device was approaching initial commercial transactions, while the financial statements still reflected a development-funded operation.[2][5]
The proposed model has an imaging-device component and a software component. The filing describes software licensing that includes maintenance, image hosting and access to algorithm updates. Hardware creates the platform at the site of care, while software is intended to provide continuing functionality and support. The investor presentation frames this as device placement plus software and services, with possible capital purchase, lease or outcomes-linked arrangements. Those are business-model routes, not reported mixes of already contracted sales.[2][4]
Pricing remains a particularly important uncertainty. The September presentation says contract structure, term and pricing are subject to negotiation and that no pricing has been finalized. Its recurring-revenue model is therefore a framework for potential economics. Actual revenue per installation could vary by customer, geography, site of service, purchasing method and support scope. Until contracts and recognized revenue establish those economics, a numerical sales projection would require assumptions the company has not fixed publicly in that presentation.[4]
The initial U.S. pathway includes BARDA-supported procurement. That can lower an early customer’s acquisition barrier, but a subsidized placement and an unsubsidized purchase test different aspects of demand. The former can establish clinical access and workflow experience; the latter more directly shows willingness to pay under ordinary procurement economics. Over time, both utilization and the proportion of deployments outside the funded program would help describe the durability of the commercial model.[4]
Emergency departments are a priority market in the October 13 release discussing the company’s participation at the ACEP Scientific Assembly, held October 5–8 in Chicago. Management highlighted the challenge of limited access to specialist burn expertise and the opportunity for an adjunctive assessment tool. Participation and product display show outreach to a relevant clinical audience. They do not by themselves report an order, a paid installation, a procurement commitment or improved outcomes in routine use.[8]
A sales organization also creates costs before revenue. The August results release reported higher sales and marketing activity ahead of first commercial sales, including a third-party pricing study. A disciplined commercial transition would connect this spending to a measurable customer funnel, signed agreements, accepted installations and sustained use. Without that connection, a larger sales effort can increase cash consumption without yet proving customer economics.[5]
On October 6, Spectral AI announced submission of an NTAP application to the Centers for Medicare & Medicaid Services for DeepView using the alternative pathway. The company describes NTAP as an additional hospital inpatient payment for eligible new technologies or services meeting CMS requirements. This is a payment mechanism associated with qualifying inpatient cases; it is not a universal reimbursement promise covering every hospital encounter or every site of care.[6]
Management anticipates a final CMS determination in August 2027. If approved, the company says NTAP status would generally become available beginning in October 2027 and continue for approximately three years. Each part of that timeline remains conditional. The application has been submitted; the expected determination is in the future; and the possible payment start follows only if the relevant CMS decision and requirements support it. The timeline does not provide additional NTAP receipts during the planned initial 2026 sales period.[6]
The release says additional payment is generally limited to approximately 65% of the estimated cost of the technology, subject to CMS requirements and payment limitations. That is a description of program mechanics, not 65% of Spectral AI’s revenue, not a reimbursement rate established for each DeepView installation, and not a guarantee that a hospital can recover its entire purchase price. The final commercial effect would depend on eligibility, utilization, case economics and the terms of any eventual determination.[6]
The company links its Breakthrough Device Designation and De Novo marketing authorization to its pursuit of reimbursement opportunities. Those regulatory milestones and the payment application are related in the strategy but remain institutionally distinct. FDA authorizes a medical device for its intended use; CMS determines applicable Medicare payment treatment. A granted De Novo request does not automatically resolve the reimbursement question, just as an inpatient payment pathway does not expand the authorized label.[3][6]
The September presentation also describes work on outpatient software-as-a-medical-service treatment, a request for new-technology APC assignment and a Category III code. These are management’s reported reimbursement initiatives and proposals. The operational question is the final payment environment in each site of service, rather than the existence of a code or an application alone. The clearly dated October NTAP announcement supplies the most specific forward timetable for that separate inpatient application.[4][6]
The DeepView System received its initial UKCA authorization for the burn indication in February 2024, according to the 10-Q and current releases. The August results release still described a planned UKCA authorization label update to reflect the FDA-cleared system for sales in the UK, Australia or Gulf Cooperation Council countries. The initial UK mark, the updated configuration or label work, and future sales in particular countries are separate milestones. U.S. De Novo authorization is not an automatic authorization in every overseas market.[2][5]
The annual report describes a July 2024 memorandum of understanding with PolyNovo to support expansion in Australia through the Special Access Scheme. As of December 31, 2025, devices had been delivered to three hospitals in Melbourne, Perth and Sydney under that pathway. Those deliveries provide access and collaboration history. The second-quarter filing’s statement that no product revenue had been generated remains the financial boundary: physical deployment for research or special access did not establish a record of commercial product sales.[1][2]
The head, hands and feet study is based at UK sites and seeks evidence for a potential label expansion. UK research activity can contribute to the development program without representing a completed regulatory expansion or a new paid market. Country-level clinical access, authorization, commercial distribution and recognized revenue need to be assessed separately, especially when a global product plan uses several of these routes at once.[7]
The handheld DeepView program is another distinct development line. The 10-Q discusses MTEC and Defense Health Agency awards supporting a battlefield-oriented version of the device. The August results release says the company completed remaining milestones under its Department of Defense contract through MTEC, including delivery of a fully functioning prototype. A prototype delivery is a meaningful engineering and contractual milestone. It is not evidence of a broad military procurement program or completed commercial handheld rollout.[2][5]
The September presentation lists further handheld program selection as an expected late-2026 milestone. That is a company expectation rather than a funded award already won. The product’s compact form could create different use cases, but development, field validation, program selection, procurement and product revenue remain distinct steps. Keeping those steps separate prevents the handheld opportunity from obscuring the nearer commercial test of the authorized cart-based burn system.[4][5]
For 2025, the latest annual report records research and development revenue of $19.650 million, down from $29.581 million in 2024. Gross profit was $8.925 million, compared with $13.274 million. The revenue decline reflected the changing phase and volume of reimbursed work under the government development program. That history shows why lower revenue can coincide with advancement toward an authorization milestone: a cost-reimbursed development contract does not behave like a recurring commercial sales franchise.[1]
The 2025 operating loss was $8.603 million, compared with $6.582 million in 2024. The latest 10-K reports a 2025 net loss of $7.571 million and a revised comparative 2024 net loss of $15.158 million. The annual filing explains a $157,000 adjustment to the 2024 income tax provision. The newer annual statement therefore provides the comparative figure for financial analysis, rather than the earlier 2024 amount repeated in the March earnings release.[1]
The difference between operating loss and net loss is also material. Fair-value changes in warrant liabilities and other financing items can create gains or expenses without reflecting hospital demand or device margins. In 2025, the company reported a $3.249 million favorable change in warrant liability fair value. That benefit affected net loss while leaving the underlying operating loss and the cash requirements of the business to be assessed separately.[1]
Cash at December 31, 2025 was $15.394 million, compared with $5.157 million at December 31, 2024. The year-end increase was supported by financing activity, not positive operating cash flow. Net cash used in operations during 2025 was $9.920 million, while financing activities provided $20.120 million. Ending cash can therefore improve during a loss-making development period when equity issuance, debt draws and warrant or option exercises more than offset operating consumption.[1]
Management’s annual liquidity assessment combined cash with government contracts, financing arrangements and certain cost-saving measures to support at least a year of operations beyond release of the financial statements. That is a conditional operating-plan assessment, not a claim that year-end cash alone covered every future requirement. The subsequent quarterly assessment uses an updated set of resources and assumptions. Neither assessment converts possible future financing into cash already held.[1][2]
Second-quarter 2026 revenue was $3.524 million, compared with $5.065 million in the corresponding 2025 quarter. First-half revenue was $7.515 million, compared with $11.772 million. All of that reported revenue remained research and development revenue. BARDA supplied $3.279 million in Q2 and $6.589 million in the first half, while other U.S. governmental authorities supplied $0.245 million and $0.926 million respectively.[2]
The Q2 year-over-year revenue decline was approximately 30.4%; the first-half decline was approximately 36.2%. Sequentially, subtracting Q2 from the first-half total gives calculated Q1 revenue of $3.991 million. Q2 was therefore approximately 11.7% lower than that calculated Q1 amount. This sequential figure is derived from the half-year and quarterly totals in the same filing; it is not a separate observation that adds independent corroboration to those totals.[2]
Q2 gross profit was $1.113 million on $3.524 million of revenue, producing approximately 31.6% gross margin. First-half gross profit was $3.140 million, or approximately 41.8% of revenue. The corresponding 2025 margins were approximately 45.2% and 46.4%. The company attributes the deterioration to cost-sharing under follow-on BARDA work and a lower realized margin on MTEC activity. Margin during the development phase reflects reimbursement and project mix, not established device-selling economics.[2][5]
Operating expenses rose to $5.437 million in Q2 from $4.413 million a year earlier. The first-half total was $9.435 million, compared with $8.477 million. Q2 expenses comprised $1.696 million of research and development, $3.088 million of general and administrative expense and $0.653 million of selling and marketing. The company changed its presentation beginning April 1, separating certain costs formerly grouped in G&A and applying the classification change retrospectively to the comparative periods.[2]
Q2 operating loss was $4.324 million, while net loss was $4.176 million, or $0.13 per basic and diluted share. First-half operating loss was $6.295 million and net loss was $7.588 million, or $0.24 per share. Although Q2 net loss was smaller than the prior year’s $7.968 million, the operating loss widened from $2.123 million. The apparent improvement in net loss was heavily affected by warrant fair-value movement, rather than an improvement in the core operating result.[2]
The August release reported adjusted EBITDA of approximately negative $3.5 million for Q2 and negative $5.2 million for the first half. This is a non-GAAP measure that excludes taxes, depreciation, net interest and several other adjustments, including stock compensation and warrant remeasurement. It can help explain management’s operating-performance view, but it excludes recurring costs and is not a measure of liquidity. The GAAP losses and cash-flow statement remain necessary alongside it.[5]
Cash and cash equivalents decreased from $15.394 million at December 31, 2025 to $13.980 million at June 30, 2026. The decrease of $1.414 million is much smaller than the first half’s $7.825 million of operating cash use because financing provided $6.422 million. A negative $0.011 million currency effect completes the bridge: $15.394 million less $7.825 million plus $6.422 million less $0.011 million equals $13.980 million. The ending balance therefore reflects both operating consumption and financing support.[2]
Operating cash use was approximately $1.304 million per month when the six-month total is divided by six. Dividing June cash by that historical monthly average produces approximately 10.7 months. This is a cash-only arithmetic sensitivity, not management guidance or a forecast of the exhaustion date. Commercial preparation, development cost share, debt payments, working capital and contract receipts can all alter the future rate. It cannot be equated with the company’s broader operating-plan liquidity statement.[2]
Management’s June assessment said cash and equivalents, PBS BARDA commitments and availability under the Yorkville standby equity agreement could fund operations for at least one year beyond release of the interim statements. The filing also warns that assumptions may be wrong, capital could be used sooner, and additional financing might be required on unfavorable terms or be unavailable. The assessment is conditional on several resources, including equity capacity; it is not a cash-only promise of a particular number of months.[2]
The June balance sheet shows $3.859 million of current notes payable and $11.055 million of long-term notes payable, for total carrying value of $14.914 million. Notes payable less cash gives calculated net debt of $0.934 million on a carrying-value basis. Lease liabilities of $1.342 million are separate from that calculation. The debt note also reports $15.859 million of gross notes and final-payment obligations before $0.945 million of unamortized discounts and issuance costs. Gross contractual obligations and balance-sheet carrying values answer different questions.[2]
Current assets were $17.792 million against current liabilities of $19.869 million, producing a calculated working-capital deficit of $2.077 million. Warrant liabilities accounted for $11.780 million of current liabilities. Those derivative liabilities are not ordinary supplier invoices, but they still belong in the GAAP balance-sheet total. Their inclusion helps explain why a working-capital deficit and a meaningful cash balance coexist. Excluding them would produce a different measure that would require its own clear definition.[2]
Total stockholders’ deficit was $12.054 million at June 30, with accumulated deficit of $63.388 million. These are accounting measures of accumulated losses and equity, not a direct measure of clinical utility or liquidation proceeds. They nevertheless underscore that the company has consumed more capital than its retained earnings have generated. A successful product launch would need to improve the financial trajectory over time, not merely increase awareness of the device.[2]
The Avenue Capital financing permitted total borrowing of $15.0 million. Spectral AI initially drew $8.5 million in March 2025. On June 18, 2026 it drew the second $6.5 million tranche after achieving the required funding conditions. Those conditions included FDA clearance and a $7.0 million equity raise. The $6.5 million belongs in the first-half financing cash flows and the June cash balance; adding it again to June cash would double-count an inflow already received.[2]
The facility matures March 1, 2028. Its interest rate is the greater of prime plus 5.25% or 12.75%, and the interest-only period was extended to 24 months after the relevant milestone achievement. A final payment of approximately $0.8 million is also due at maturity. The debt note’s exact principal-plus-final-payment amount is $15.750 million. These terms make the borrowing a real cost of the commercialization bridge, rather than a permanent source of non-dilutive capital with no future claim.[2]
Up to $2.0 million of the Avenue borrowing can be converted at the lender’s option at 120% of the Avenue warrant exercise price. The financing also included 768,072 warrants, with a $1.66 exercise price at June 30. Conversion and warrant features mean the instrument has potential equity consequences alongside interest and repayment obligations. The company’s August 11 results release describes the draw as non-dilutive capital.[5] The full financing package includes those potential rights and must be understood in that broader form.[2]
Debt can extend the interval available for execution without requiring an immediate public share sale. It also creates a later repayment burden and recurring financing costs. Whether that trade-off ultimately benefits existing holders depends on the operating progress financed by it and on the capital needed before maturity. The same balance sheet can therefore show improved near-term liquidity and greater future obligations. The cash-flow bridge and the gross-debt table make both effects visible.[2]
Common shares outstanding increased from 30,688,895 at December 31, 2025 to 32,184,928 at June 30, 2026. The increase was 1,496,033 shares, or approximately 4.9%. The statement of changes in equity identifies pre-funded warrant exercises, employee and director award vesting, option exercises, a small Yorkville issuance and investor warrant exercises among the components. Potential issuance capacity and shares already issued are therefore different parts of the capital story.[2]
During the first half, 935,000 pre-funded warrants were exercised, leaving none outstanding at June 30. Their exercise price was $0.0001 per share. Most of their purchase consideration had been associated with the earlier financing, so the conversion into common shares is not equivalent to receiving the original purchase price again in 2026. The filing separately reports 30,000 investor warrant exercises and 3,000 shares sold through the Yorkville agreement for approximately $5,000.[2]
At June 30, the company reported 8,433,333 public warrants with a $2.75 exercise price, 635,000 investor warrants at $1.80, 768,072 Avenue warrants at $1.66 and 4,000,000 Hudson warrants at $2.51. It also reported 77,017 Angel warrants at $7.32 in that dated balance-sheet disclosure. These instruments have different contractual terms; the historical outstanding counts are not a claim that every warrant remains unchanged indefinitely after the balance-sheet date.[2]
The loss-per-share note excluded 19,246,901 potentially dilutive instruments from diluted EPS because their inclusion would have been anti-dilutive in a loss period. The total includes options, warrants, unvested restricted stock and convertible-note shares. It is not an issued-share count and it is not a forecast that all instruments will exercise. A loss-period diluted EPS denominator can equal basic EPS while meaningful potential share issuance still exists.[2]
Yorkville’s standby equity purchase agreement expires in March 2027. At June 30, 2,791,113 of 6,369,937 registered resale shares remained available for issuance. The agreement prices shares at a discount to market and imposes ownership, registration and advance conditions. The available shares do not represent a fixed amount of cash. Their dollar value changes with the share price and with the company’s ability to satisfy the agreement’s requirements.[2]
The quarterly filing also describes an effective $50.0 million shelf, with approximately $12.1 million utilized at June 30. The gross difference is not freely usable capacity because the company’s non-affiliate public float was below the $75.0 million threshold. The one-third-public-float limitation and preceding sales reduced available capacity to approximately $10.9 million on the dates described in the filing. That amount is variable, not cash in the bank and not an amount guaranteed to be raised.[2]
Spectral AI’s annual report identifies nine active patent families covering imaging, classification, wound assessment and related tissue-analysis applications. At December 31, 2025 it reported 13 issued and allowed U.S. patents and five pending U.S. applications, plus 21 issued and allowed international patents and 23 pending foreign and international applications. “Issued and allowed” is the company’s combined category; pending applications remain a separate category and are not granted patents.[1]
The company also reported 47 trademarks and 11 pending trademark applications. Patents, trade secrets, proprietary data and trademarks protect different aspects of the business. Trademarks support product identity; patents address specified inventions; confidential algorithms and data can support technical differentiation. The annual risk disclosures acknowledge that patents can be challenged, applications can fail, enforcement is costly and competitors can develop alternative approaches. A portfolio count describes assets and applications, not immunity from competition.[1]
Spectral IP was formed as a wholly owned subsidiary to acquire and monetize AI intellectual property with emphasis on healthcare. On May 5, 2025, it received a worldwide non-exclusive license to one international patent asset for commercialization and monetization outside the parent’s core areas of focus. The annual report says market terms and conditions were to be finalized. A non-exclusive license to one asset is narrower than transferring the full DeepView platform or all proprietary data into a separate listed company.[1]
The subsequent corporate status changed. On July 24, 2026 Spectral IP requested withdrawal of its Form S-1 registration statement, originally filed in March 2025 for a proposed IPO. The 10-Q states that no securities were sold under the registration statement and that Spectral IP remained wholly owned and consolidated. The withdrawal produced no adjustment to the June interim statements. The proposed offering therefore cannot be treated as a completed spinout, a realized cash inflow or a distribution already delivered to MDAI holders.[2]
The withdrawal keeps the current operating analysis focused on the consolidated group. Intellectual-property monetization may still be an opportunity, but a historical proposed IPO is not a present separate public valuation. Any future transaction would have to specify the assets, rights, economic consideration and ownership effects. Until such terms exist, the observable commercial and financial tests remain DeepView adoption, development-program performance and the parent company’s capital needs.[1][2]
The annual report says manufacturing is outsourced, with Cobalt Product Solutions in Plano, Texas involved in the current-generation system and additional specialist suppliers supporting optics, design and electronics. Spectral AI had no current plan to build its own manufacturing facility. This model can avoid the capital cost of owning a full factory, but it still requires reliable suppliers, quality oversight, component availability and installation support. A cleared design becomes a commercial product only when those operations can deliver consistently.[1]
The 10-Q notes supplier concentration and the possibility of shortages, longer lead times, cost increases and shipment delays. It also anticipates spending on manufacturing scale-up and commercial personnel. Those risks can matter even at a modest initial rollout: a delayed component or installation can postpone recognized revenue while staff and development costs continue. Hardware scaling is therefore a separate operating test from the software’s diagnostic performance.[2]
The leadership group entering commercialization included CEO Vincent Capone, CFO David McGuire, COO Stan Micek and CCO Darcy Bajko, as described in the September investor presentation. The presentation places clinical, financial and commercial functions within the same launch organization. The relevant execution outcomes are not the titles alone, but the ability to move from a development-funded program to negotiated customers, repeatable manufacturing, reliable accounting and supported clinical use.[4]
Financial controls are an explicit unresolved execution issue in the June filing. Management concluded that disclosure controls and procedures were not effective because of a material weakness in financial-statement close processes, including reconciliations, transaction processing and review. The remediation plan includes accounting personnel, documentation, systems improvements and specialist support. Management expected testing during 2026 to support a future conclusion of effective operation, but the June conclusion remained ineffective.[2]
The annual auditor’s unqualified opinion on the 2025 financial statements does not erase that control disclosure. Financial-statement assurance and internal-control effectiveness are separate. For a company adding commercial contracts, inventory, support obligations and more varied revenue arrangements, better financial processes have practical importance. The combination of initial sales and reliable reporting would provide stronger evidence of execution than either a new executive appointment or a remediation plan on its own.[1][2]
Finviz’s October 13, 2026 common-stock data show a regular-session close of $1.56 and volume of 178,183 shares. The vendor reported market capitalization of approximately $50.21 million, approximately 32.18 million shares outstanding and a float of approximately 26.96 million shares. These are dated, rounded market-data fields. The SEC’s 32,184,928 common shares at June 30 are a company count at a different reference date, not proof of a simultaneous October capitalization denominator.[11][2]
The same dated Finviz data show insider ownership of 16.24%, institutional ownership of 12.06% and short interest equal to 8.70% of float, with a short ratio of 15.83. Ownership classifications and short-interest dates follow vendor conventions and may not update on the same schedule as price. They provide market context rather than a clinical or financial fact about DeepView. In particular, institutional ownership is not endorsement of product economics, and short positioning is not proof that a regulatory or commercial outcome will fail.[11]
A relatively small dollar capitalization and modest daily share volume can coexist with a material development program. They also mean that liquidity, spreads and headline sensitivity can matter independently of the operating thesis. The October volume is one day’s observation, not a guarantee that a similar order can be executed under future market conditions. MDAIW remains a separate security with separate pricing and warrant terms.[11][2]
The main business milestones do not establish whether $1.56 is an appropriate price for the shares. Revenue composition, cash needs, debt and potential dilution influence the economic picture, while the market sets the trading price. The Health Score describes financial and operational robustness, not a target price or a return forecast. A stronger launch can still coincide with financing dilution; a lower share price can still coexist with progress on a device. Those relationships prevent a single milestone from carrying more meaning than it supports.
Evidence of signed commercial agreements, accepted paid installations and product revenue would directly address the current transition. Continued utilization and software renewal would address durability. Broader unsubsidized purchasing would show whether the proposition travels beyond the government-supported initial pathway. Management’s stated device-and-software model makes these distinct tests relevant: selling a device, using it routinely and maintaining a recurring relationship are different events.[2][4]
The cautious interpretation would weaken if commercial revenue began to cover a meaningful share of operating spending while cash use moderated without repeated equity issuance. It would strengthen if first sales slipped materially beyond the company target, commercial costs rose faster than placements, or the government development stream narrowed without an adequate replacement. These are conditional operating scenarios, not predictions that a particular outcome will occur.[2][5]
Clinical interpretation would change with evidence from the new anatomical study and the planned outcomes work. A completed validated dataset can address performance within an expanded population; a study demonstrating changed treatment or resource use can address adoption economics. Those findings would remain specific to the study design, population, endpoints and authorized indication. The original UK observational paper’s call for more outcomes and workflow evidence is an important boundary on how far present diagnostic performance can be generalized.[7][9]
Financing interpretation would change with actual cash movements and contractual terms. Government options being exercised would increase committed program scope, but the related revenue would still depend on performance. An equity issuance would add cash and shares; a debt repayment would reduce both cash and obligations. The current cash-only historical sensitivity and management’s resource-based liquidity assessment would need to evolve with those events rather than remain fixed descriptions of future capacity.[2]
The completed milestones include initial UKCA authorization in February 2024, FDA De Novo classification dated May 21, 2026, the second Avenue tranche drawn on June 18 and withdrawal of the proposed Spectral IP offering requested on July 24.[2][3] The August results release described completed handheld prototype delivery under MTEC.[5] The October 1 release added first site initiation for head, hands and feet research, and the October 6 release added the submitted NTAP application.[6][7] Each event advances a particular layer of the program, not all layers at once.
The latest October 13 announcement reported DeepView’s display at the ACEP Scientific Assembly that had already taken place October 5–8. The September 24 ACS announcement concerned a planned presentation during the September 26–29 congress. Those meetings are past events, not upcoming catalysts. The ACEP release confirms participation; the earlier ACS announcement records the scheduled presentation and does not itself supply a later outcome report.[8][10]
Nearer company objectives include first U.S. commercial sales by year-end 2026 and initiation of the Triage and Treatment Outcome Study in Q4 2026.[5] The September presentation also described planned BARDA-supported deployment through June 2027 and further handheld selection expectations. These are company timelines with execution dependencies, not fixed regulatory decision dates. The October label study is expected to complete in 2027, while the NTAP release anticipates an August 2027 CMS determination and a conditional October 2027 payment start.[4][6][7]
The central operating question is now the connection between those milestones and the financial statements. FDA authorization establishes the opportunity to market the specified device. Government commitments support development and procurement work. Research can broaden the evidence. Commercial transactions and repeated use would establish a different kind of proof: that the platform can function as a sustainable healthcare business. Cash consumption, financing obligations and per-share dilution determine how much time and capital are available to reach that stage.[2][3][4]
These are conditional operating tests, not forecasts. Their factual premises are the company’s stated launch objectives, financing position and research programs.[2][5][7]
Yes. FDA granted De Novo classification through DEN250028 on May 21, 2026 for the specified adjunctive thermal-burn assessment use. It is not a standalone diagnostic or a definitive excision-border tool.[3]
No. It is the approximate maximum contract value. The June filing reports $86.6 million committed and $63.4 million of further options, plus a $9.7 million company development contribution. Revenue follows work performed.[2]
No product revenue was reported in the second-quarter filing. Management targeted first U.S. commercial sales by year-end 2026.[2][5]
The October 6 announcement reports an application. Management expects a CMS determination in August 2027 and a conditional payment start in October 2027.[6]
The July 24 withdrawal request related to its proposed IPO. No securities were sold under that registration; the subsidiary remained wholly owned and consolidated.[2]
No. MDAI is common stock; MDAIW is the separately listed redeemable warrant. The filing lists distinct classes and terms.[2]
Every Merlintrader stock hub, catalyst update and market brief is published to Telegram the moment it goes live. No paywall, no spam, just the research.
Disclaimer. This content is published by Merlintrader for educational and informational purposes only. It is independent journalism and research. It does not constitute investment advice, an investment recommendation, an offer or a solicitation to buy or sell any security, and it is not a research report within the meaning of applicable United States securities regulation. Nothing here should be read as a recommendation to buy, sell or hold $MDAI or any other security.
Figures are taken from public filings with the U.S. Securities and Exchange Commission, company press releases and market-data providers, and are stated with their reference dates. Data can change without notice, and figures published before a results release become outdated the moment that release is issued. Merlintrader makes no representation that the information is complete or current at the time of reading. Readers should verify every figure against the primary source before acting on it.
Medical-device companies carry clinical, regulatory, reimbursement, adoption, manufacturing and financing risks. Government funding and hospital purchasing can be delayed, research may not support expanded uses, and commercial deployment can consume cash before revenue grows. Authorization and contract announcements do not guarantee profitable sales or cash collection. Additional borrowing or equity issuance can increase obligations or dilute existing shareholders. Investors can lose part or all of their capital. Every reader is responsible for their own decisions and should consult a licensed financial adviser where appropriate.
Merlintrader may hold positions in securities mentioned. Some links on this page are affiliate or referral links, including those to Finviz and Stocktwits, which may generate a commission at no cost to the reader. Full legal information is available on the disclaimer and terms of use and privacy pages.
Regulatory and clinical dates across healthcare, with their stated sources and dependencies.
Open the free catalyst calendar →
