RadNet: The AI Story That Doesn’t Add Up – Bethany McLean
RadNet's AI transformation exposed: Short report reveals accounting manipulations, artificial growth, and valuation disconnect in medical imaging company, challenging inflated tech narrative and market positioning.
HunterBrook Capital alleges that RadNet's transformation into an "AI-powered" healthcare company is largely smoke and mirrors, a strategic rebranding of a mature, low-margin imaging center roll-up that uses accounting maneuvers, center consolidations, and aggressive financial adjustments to manufacture growth and justify a valuation nearly double that of comparable peers.
Ticker: RDNT (NASDAQ)
Position Disclosure: HunterBrook Capital discloses that they hold a short position in RadNet and stand to benefit if the stock price declines.
Report Author: HunterBrook Capital
Original Report: Read the full analysis here
Core Thesis
RadNet's problems center on:
- Artificial Growth Through Consolidation: Up to 57% of RadNet's reported 6% same-center revenue growth may stem from consolidating nearby imaging centers and redirecting patient volume rather than genuine organic demand. True underlying growth appears closer to 2.5–3% annually.
- Overstated AI Pivot: Digital Health/AI operations generate less than 5% of total revenue (~$65 million of $1.5 billion), with the majority coming from internal sales to RadNet's own imaging centers rather than validated external customers, calling into question the substance behind the "AI company" narrative.
- Accounting Opacity and Inconsistent Disclosures: SEC filings contain material discrepancies in center counts (398 vs. 375 locations; 44 new centers vs. 9), conflicting revenue figures, and unexplained reclassifications (e.g., eRad moved into Digital Health), making it difficult to verify actual footprint expansion or contraction.
- Aggressive Adjusted Metrics Mask Margin Decline: Management presents "Adjusted EBITDA" that excludes stock-based compensation and key R&D expenses. When these costs are included ("Corrected EBITDA"), operating margins appear to be shrinking, not expanding, contradicting the growth story.
- Significant Insider Selling with No Buying: Over 780,000 shares (~$50.9 million) sold by insiders over two years, including substantial sales by CFO Mark Stolper, with zero corresponding open-market purchases, raising questions about management confidence in the AI transformation thesis.
- Valuation Disconnect: RadNet trades at ~19x EBITDA (or ~30x "Corrected EBITDA"), nearly double comparable imaging peer Lumexa (10–12x at IPO). The implied Digital Health valuation exceeds $3 billion (~40x revenue), despite minimal external traction and unproven scalability.
- Questionable Governance Practices: Corporate expenses include $180,000 annually for a New York apartment tied to the CEO's family trust (despite California headquarters), overlapping brand structures, and intercompany transactions that blur operational reality.
- Unproven AI Acquisitions: Recent purchases of iCAD ($100 million) and DeepHealth carry lofty valuations with unclear integration benefits, scalable synergies, or material margin improvement, appearing more like branding exercises than transformative technology plays.
Notable Findings & Intriguing Details
HunterBrook Capital's investigation uncovered several eye-opening details:
- The "AI Read" Commodity: RadNet charges patients an additional $40 for an AI-enhanced mammogram review, but industry experts and radiologists characterize this as a commoditized service available from multiple competitors at similar or higher prices, hardly a proprietary innovation.
- Center Consolidation Detective Work: Using Internet Archive's Wayback Machine, researchers documented specific cases (Phoenix, Bakersfield) where RadNet closed locations and redirected patients to nearby centers, mechanically boosting "same-center" revenue without actual organic growth.
- Branding Maze: Approximately 100 of RadNet's 400 centers sit within a quarter-mile of another RadNet location, with multiple entities (Breastlink, Arizona Diagnostic Radiology) operating at shared addresses, creating attribution confusion and suggesting substantial overlap rather than true expansion.
- Radiologist Skepticism: External radiologists, including Dr. Roman Keller, and AI expert Andrej Karpathy's observations cast doubt on AI's near-term transformative potential in high-risk, complex radiology, contradicting RadNet's aggressive AI positioning.
- CFO's Major Exit: CFO Mark Stolper sold a sizable portion of his personal holdings even as he publicly projected 30% annual Digital Health growth "for many years", a stark disconnect between public optimism and personal actions.
- Dot-Com Parallels: The report draws comparisons to late-1990s companies that added ".com" to their names for instant valuation boosts, suggesting RadNet's "AI" rebrand follows a similar playbook of hype over substance.
- Medicare Headwinds: RadNet's core imaging business faces ongoing reimbursement pressure from CMS cuts, making the mature brick-and-mortar network increasingly challenged on margins, precisely when the company needs the AI story to work.
- Delayed Promises: Management projections that DeepHealth would be fully integrated by year-end and externally licensed by 2025 have shifted into vague, underdelivered timelines, suggesting operational execution lags far behind investor expectations.
Frequently Asked Questions (FAQs)
What is RadNet's core business?
RadNet operates a network of approximately 400 outpatient imaging centers providing diagnostic services like MRIs, CT scans, and mammograms. Despite recent rebranding efforts emphasizing AI and Digital Health, the company remains fundamentally a mature, brick-and-mortar radiology roll-up dependent on Medicare/insurance reimbursement.
How significant is RadNet's AI and Digital Health business?
Digital Health generated less than 5% of total revenue, approximately $65 million out of $1.5 billion in the first nine months of 2025, with the majority coming from internal sales to RadNet's own imaging centers rather than external customers, raising questions about the division's genuine market traction.
What is "same-center revenue growth" and why does it matter?
Same-center growth measures revenue increases at locations open for comparable periods, excluding new acquisitions. The report alleges RadNet artificially inflates this metric by consolidating nearby centers and redirecting patient volume, with modeling suggesting up to 57% of reported 6% growth may be mechanical rather than organic, implying true underlying growth closer to 2.5–3%.
What accounting issues does the report identify?
Material inconsistencies across SEC filings: conflicting center counts (398 vs. 375), discrepancies in new center additions (44 vs. 9), unexplained revenue figure variations, and reclassifications (eRad into Digital Health) that obscure operational reality. The report also criticizes "Adjusted EBITDA" that excludes stock-based compensation and R&D, masking declining margins when these costs are included.
What is the significance of insider selling at RadNet?
Insiders sold over 780,000 shares (~$50.9 million) over two years with zero open-market purchases, including substantial sales by CFO Mark Stolper. HunterBrook Capital argues this pattern, particularly while management publicly touts aggressive Digital Health growth projections, raises questions about leadership confidence in the AI transformation thesis.
How does RadNet's valuation compare to peers?
RadNet trades at approximately 19x EBITDA (or ~30x when adjusted for excluded costs), nearly double imaging peer Lumexa's 10–12x IPO multiple. HunterBrook Capital calculates that the implied Digital Health valuation exceeds $3 billion (~40x revenue), which appears disconnected from the division's minimal external revenue and unproven scalability.
What are the main concerns about RadNet's AI acquisitions?
Questions whether acquisitions like iCAD ($100 million) and DeepHealth deliver genuine value, noting that durable monetization, scalable synergies, and material margin improvement remain unproven. The report suggests these purchases function more as branding exercises to support the AI narrative than transformative technology investments with demonstrated ROI.
What governance red flags does the report highlight?
Several unusual practices: $180,000 annual rent paid by the company for a New York apartment tied to the CEO's family trust (despite California headquarters), overlapping brand structures creating attribution confusion, complex intercompany transactions, and inconsistent disclosure practices that obscure operational clarity.
What is the "AI read" service and why is it questioned?
RadNet charges patients an additional $40 for AI-enhanced mammogram analysis through its Digital Health platform. However, HunterBrook Capital's industry expert consultations characterize this as a commoditized service available from multiple competitors at similar prices, undermining claims of proprietary innovation or sustainable competitive advantage.
What do radiologists think about AI in radiology?
The report cites skepticism from practicing radiologists, including Dr. Roman Keller, and references AI expert Andrej Karpathy's observations about AI's limited near-term applicability in complex, high-liability medical specialties like radiology, contradicting RadNet's positioning of AI as an imminent transformative force in its business.
How did HunterBrook Capital verify center consolidations?
They used Internet Archive's Wayback Machine to track RadNet's online center locator over time, documenting specific instances (Phoenix, Bakersfield) where locations disappeared from listings while nearby centers remained, suggesting closures that mechanically redirect patient volume to surviving sites and artificially boost same-center growth metrics.
What are the Medicare reimbursement headwinds?
RadNet's core imaging business faces ongoing CMS (Centers for Medicare & Medicaid Services) reimbursement cuts, pressuring margins in the mature brick-and-mortar network, creating urgency for the AI/Digital Health pivot to succeed while simultaneously making the core business increasingly challenged.
Has RadNet management delivered on AI integration timelines?
Management projections that DeepHealth would be fully integrated into workflows by year-end and externally licensed by 2025 have shifted into vague, underdelivered plans, suggesting execution lags significantly behind investor expectations and calling into question the feasibility of aggressive growth targets.
What is HunterBrook Capital's position in RadNet?
HunterBrook Capital explicitly discloses holding a short position in RadNet stock (ticker: RDNT), meaning they stand to profit financially if the share price declines, a material conflict of interest readers should consider when evaluating the research conclusions.
Did RadNet respond to these allegations?
RadNet did not provide direct comment on the research findings; external inquiries were directed back to HunterBrook for RadNet's perspective, leaving the company's specific rebuttals to the allegations unclear.
What is the bottom-line investment implication?
RadNet's AI rebranding appears to be a strategic narrative designed to justify premium valuations on a mature, low-growth imaging business facing reimbursement pressure, with reported growth metrics artificially inflated through center consolidations and accounting adjustments rather than genuine operational transformation, suggesting significant downside risk to current valuation levels.
Important Disclosure
This summary is based entirely on third-party research conducted by HunterBrook Capital and does not constitute original analysis. The full, detailed report, including extensive supporting evidence, data modeling, and specific examples, is available at https://hntrbrk.com/radnet/.
HunterBrook Capital explicitly discloses that they hold a short position in RadNet, Inc. (RDNT) and stand to benefit financially if the stock price declines. Readers should consider this material conflict of interest when evaluating the research conclusions and conduct their own independent due diligence.
All findings, allegations, and conclusions presented in this summary are attributed to HunterBrook Capital's original research and analysis. This summary is intended for informational purposes only and does not constitute investment advice.
Original Research Credit: HunterBrook Capital
Full Report: https://hntrbrk.com/radnet/
Disclaimer
ShortReport.fyi is not the author or originator of the content provided in this publication. We act solely as a distributor of the content and do not endorse, verify, or take responsibility for the accuracy, completeness, or reliability of the information presented. Each author retains full ownership and responsibility for their respective content, including any opinions, projections, or analyses expressed therein.
This publication is for informational purposes only and should not be construed as legal, business, investment, or tax advice. Readers are advised to consult the original source material and seek guidance from qualified professionals before making any decisions based on the information contained herein.
ShortReport.fyi disclaims all liability for any loss or damage arising from reliance on the content provided by third parties. The views expressed in these materials are solely those of the respective authors and do not necessarily reflect the views or opinions of ShortReport.fyi.