InterDigital, Inc.: ARR Hit a Record 625.7 Million but 103.7 Million of Q2 Revenue Was Catch-Up, and 326.08 USD Sits 42% Above the 230 USD Conservative Ceiling

The filing does not disclose an Amazon-specific downside sensitivity in dollars. That omission is consequential. If an eventual award changes the transaction price, ASC 606 requires the estimate to be updated, with the cumulative effect flowing through revenue as appropriate. A lower award can cause a negative catch-up relative to the prior estimate; a higher award can create positive catch-up. The constraint is intended to reduce the chance of a significant reversal, not eliminate it. Contract assets increased sharply by June 2026, consistent with more revenue having been recognised ahead of cash billing or receipt, although the filing does not identify the entire increase as Amazon.

A useful decomposition of the new 2026 guidance follows from the company's own numbers. Q1 revenue was 205.4 million with 63.6 million of catch-up, leaving about 141.8 million recurring. Q2 contributed 156.4 million recurring and 103.7 million catch-up. If second-half recurring revenue merely stays near the Q2 run rate, full-year recurring revenue would be roughly 611 million. Against the 775–845 million guide, that implies approximately 164–234 million of full-year catch-up or revenue above that flat recurring run rate. The first half has already produced 167.3 million of catch-up. On that simple sensitivity, the low end needs almost no additional second-half catch-up; the midpoint implies about another 32 million; the high end requires about another 67 million. This is my arithmetic, not company guidance, and it shows that the raised outlook has a much stronger recurring underpinning than simply reading the 85 million midpoint increase might suggest.

The moat is real, but it is unusual. InterDigital's protection comes from standards contributions, portfolio breadth, a balance sheet capable of financing long disputes, and litigation expertise across jurisdictions. Customer affection contributes little. A handset or streaming company can regard InterDigital as an adversary and still ultimately owe royalties. The economic equivalent of switching costs comes from the difficulty of designing standardized functionality around a sufficiently broad portfolio. This is why litigation belongs inside normal operating analysis instead of an "exceptional items" bucket.

Europe has materially strengthened this bargaining position. On June 16, 2026 the Mannheim Local Division of the UPC granted InterDigital an injunction against Disney over HEVC technology spanning eleven EU countries and upheld the patent's validity; Düsseldorf granted a second eleven-country HEVC injunction on July 23. A third UPC injunction followed on September 2, extending the pressure into another video functionality. Disney can appeal. The practical consequence has already gone beyond legal theory: Disney altered or disabled premium video features in affected markets while seeking workarounds.

The structural case for European enforcement is stronger than it was a few years ago, but extrapolating first-instance UPC decisions into a permanent patent-holder regime would be aggressive. The European Commission's proposed SEP Regulation was withdrawn, removing one prospective layer of mandatory administrative rate-setting. U.S. policy has also shifted in a more patent-holder-friendly direction: in December 2025 the USPTO formed an SEP Working Group explicitly focused on stronger, more predictable remedies, including greater recognition of injunctive and ITC relief. China remains a counterweight. The European Union challenged Chinese worldwide SEP-rate-setting practices at the WTO, while China announced in April 2026 that it would withdraw its anti-suit-injunction policy following a WTO ruling. Global FRAND jurisdiction remains contested.

The bear case has unusually concrete evidence as well. In the public Lenovo UK rate-setting dispute, the court-set rate landed substantially below the position InterDigital had advanced. The first-instance rate was about 0.175 per unit; the Court of Appeal raised it to about 0.225, still roughly 45% of the approximately 0.498-per-unit position associated with InterDigital's trial case. Comparability is imperfect because the offers, portfolio scope and time periods are complex, but the visible public precedent does not support assuming that InterDigital's opening demands routinely become judicial rates. Conversely, the Samsung arbitration materially increased annual recurring economics compared with Samsung's previous licence, so private outcomes can be much stronger. Opening arbitration positions are confidential, preventing a reliable award-to-demand base-rate study.

The balance sheet is a genuine advantage. June 30 cash, restricted cash and short-term investments were around 1.1 billion. Against that sits roughly 380 million of 3.5% 2027 convertible notes. The conversion structure needs careful treatment: the purchased note hedge substantially offsets conversion dilution above the note conversion price, while sold warrants reintroduce dilution above a 105.43 strike. The warrants cover about six million shares and settle on a net-share basis over maturities running from September 2027 through April 2028. At 326.08, the six-million headline figure does not imply six million new shares. Net economic warrant dilution is approximately

6.0m × (1 − 105.43 / 326.08) = 4.06m shares.

Adding that to roughly 25.81 million basic shares gives 29.87 million shares before employee equity awards, a 15.7% increase in the basic count. The company's Q2 non-GAAP EPS arithmetic implies a diluted weighted share base near 30.6 million once ordinary equity compensation is also reflected. At the current price, basic capitalization is about 8.42 billion; warrant-only economic capitalization is about 9.74 billion; a roughly 30.6 million fully diluted proxy is about 9.98 billion. The difference is too large to leave out of valuation.

The qualitative portrait is a re-rating: a mature patent cash generator that is broadening its royalty base and gaining stronger enforcement tools, while the share price increasingly capitalises that expansion before several of the decisive rates have become final.

The central disagreement follows. Bulls see a portfolio whose recurring base can move from 625.7 million toward management's longer-term billion-dollar scale through streaming, video, smartphones, auto and the next standards generation. Bears see a company valued at roughly 38 times trailing earnings, with around 16–19% economic dilution beyond the basic share count, where a meaningful share of recent accounting revenue rests on catch-up items, constrained estimates and legal outcomes. Both descriptions are factually defensible. The investment question is the price being paid for the second one to evolve into the first.

Company vertical history and financial review

InterDigital's modern economics make more sense once its product-company ancestry is understood. Sherwin Seligsohn founded International Mobile Machines Corporation in Philadelphia in 1972, when mobile communications meant specialized radio systems rather than standardized mass-market cellular handsets. Early IMM systems sought to connect portable radios into the public telephone network. The company began as an engineering and communications-product enterprise, not as a patent assertion vehicle. It went public as International Mobile Machines on November 12, 1981 at 3.00 per share. The company's investor FAQ confirms the date and IPO price; I could not find a sufficiently reliable primary record of the original gross proceeds, so I do not infer one from the share price.

The long corporate story divides naturally into five stages.

The first stage, from 1972 through the early digital-cellular era, was invention plus physical commercialization. IMM engineers worked on portable wireless communications well before cellular standards became the giant global coordination mechanism they are today. The legacy that matters now is institutional rather than financial: InterDigital developed a culture of producing communications inventions early enough to influence subsequent standards. The early hardware itself did not become the enduring moat.

The second stage, through the 1990s and 2000s, was a hybrid technology-development period. InterDigital worked on TDMA, CDMA and later 3G technologies, while trying several routes to commercialization, including modem and ASIC-related technology. Old SEC filings show spending on technology licences necessary for a 2G/3G dual-mode modem ASIC offering. The strategic problem was increasingly visible: competing in physical silicon required product-development cycles, manufacturing partners and commercial scale that a relatively small research company did not naturally possess, while its growing cellular patent portfolio could be licensed across the entire industry.

The third stage culminated around 2011–12, when patents became the explicit centre of gravity. The 2011 Nortel patent auction made mobile-IP portfolios highly visible strategic assets, and InterDigital launched a strategic-alternatives review amid speculation that a larger technology company might buy it. No whole-company transaction resulted. In June 2012, however, Intel agreed to pay 375 million for roughly 1,700 InterDigital patents and patent applications covering 3G, LTE and 802.11 technologies. Intel's own announcement confirms the transaction. The sale monetised only part of InterDigital's portfolio and effectively validated patents as separable financial assets.

In hindsight, this was the decisive business-model crystallisation. Hardware ambition faded and R&D remained, but R&D increasingly existed to generate licensable IP. InterDigital became financially unusual: research looked like the "factory", patent prosecution like part of maintenance capital, and litigation like a selling expense. The output did not need a manufacturing plant.

The fourth stage broadened the asset base beyond cellular. InterDigital's expansion into video accelerated through transactions involving Technicolor's patent-licensing and research activities. Those assets supplied expertise in video compression, image processing and related technologies that now underpin the consumer-electronics and streaming strategy. This diversification matters because cellular royalties ultimately depend on a fairly concentrated universe of handset manufacturers, whereas video patents can potentially monetize televisions, set-top devices, PCs, streaming devices and the streaming service itself. InterDigital now reports Smartphone, CE/IoT/Auto and Streaming and Cloud licensing programs even though GAAP reporting still treats the company as one operating segment.

The current stage began to become visible from roughly 2023. Several formerly contentious counterparties moved onto licences, arbitration strengthened Samsung economics, video enforcement became more successful, and the UPC created a new multi-country injunction venue. By early 2026 InterDigital had renewed Xiaomi, signed LG Electronics for televisions and computer displays, obtained a Brazilian decision describing its Transsion offer as FRAND, and accumulated multiple Disney injunctions. Amazon then became the first major streaming-and-cloud licence disclosed in the company's new program.

The capital market has reacted to the change in perceived durability, not merely to a single year's earnings. InterDigital's long share-price history contains repeated litigation-driven spikes and retracements. The recent leg differs because recurring revenue itself has risen alongside legal outcomes. The stock's 52-week high reached 412.60 before settling at 326.08 on September 15, 2026. Q2's large revenue beat, Amazon announcement and 85 million midpoint guidance raise produced another sharp positive reaction before some of those gains retraced.

The financial evolution reinforces that interpretation.

†Defined here as operating cash flow less property/equipment purchases and capitalized patent costs; this is the conservative owner-cash measure used later. Source: company 10-K.

Revenue was 359.0 million in 2020 and 425.4 million in 2021, and then continued upward before the much larger 2023–25 step-up. The important cause was licence economics, not unit production. New and renewed agreements, litigation catch-ups and higher negotiated rates drove the changes. In 2024, for example, large catch-up items associated with Samsung TV, OPPO and Lenovo helped take revenue to 868.5 million; 2025 revenue slipped 4% to 834.0 million even as underlying recurring licensing strengthened because the mix of catch-up revenue changed.

Margins need interpretation. A 100 million patent settlement can contribute revenue with little contemporaneous R&D expense because years of research and legal spending preceded it. The result is exceptional reported operating leverage. Conversely, the legal bill can arrive before a licence. The 55% 2025 operating margin is real accounting profit, but it is not a normal manufacturing margin that can be extrapolated by multiplying a stable gross margin through future unit growth.

Cash conversion is healthy over a multi-year period but just as lumpy. For the cleanly comparable 2023–25 period, aggregate operating cash flow was about 1.03 billion against 976 million of aggregate net income, a 1.05× conversion ratio. After treating all property/equipment spending and capitalized patent prosecution costs as owner-maintenance outlays, aggregate FCF was about 856 million, or 0.88× net income. The broader 2021–25 filings also show aggregate operating cash comfortably exceeding aggregate reported profit; timing of licence receipts drives large year-to-year differences.

2025 illustrates why a single FCF year can be misleading. Operating cash flow jumped to 544.5 million, helped by a 118.5 million reduction in receivables. Owner-style FCF approached 474 million. In the first half of 2026, by contrast, operating cash flow was only about 98.6 million and capitalized patent plus property spending about 31 million, leaving approximately 67.6 million. Nothing about the underlying patent portfolio changed enough to justify treating 474 million and 68 million as two different business qualities; contract billing, collections and legal costs moved the cash calendar.

The balance sheet gives InterDigital room to tolerate that variability. At June 30, 2026 it held approximately 1.1 billion in cash, restricted cash and short-term investments. Deferred revenue was about 442 million, meaning cash has also been collected in advance of future revenue recognition on some licences. The company disclosed around 1.9 billion of future cash payments expected under contracted fixed-price agreements, including constrained minimum estimates associated with Amazon and Lenovo. Those future receipts should not be treated as cash on the balance sheet, but they reduce financing risk.

The company simultaneously returned capital. In 2023 it spent roughly 340 million repurchasing stock, followed by 66.7 million in 2024 and 102.3 million in 2025. Q2 2026 added roughly 23 million of repurchases and about 18 million of dividends. On September 9 the board raised the quarterly dividend from 0.70 to 0.75 per share, making the annualized cash dividend 3.00. At 326.08 that is a yield of only about 0.92%, so dividends are a secondary component of the equity case.

The repurchases have reduced the basic share count, but the convertible call-spread structure complicates the per-share result. At year-end 2025 basic weighted shares were 25.8 million while diluted weighted shares reached 34.5 million under GAAP because the accounting calculation does not give full credit to economically offsetting hedge instruments. Investors need an economic dilution calculation, not either headline count in isolation.

The 2027 notes were issued with a conversion price around 76.72, together with purchased convertible-note hedges and sold warrants at 105.43. Economically, the purchased hedge protects much of the conversion spread until the warrant leg takes over. The warrant exposure covers six million shares. At 326.08, approximately 4.06 million net shares represent the warrant's intrinsic-value dilution. The company's own sensitivity table gives roughly 4.05 million incremental net shares around a 325 share price, confirming the calculation. The warrants mature through a series of net-share settlements between September 2027 and April 2028.

The 23% "six million warrants divided by 25.8 million basic shares" headline overstates immediate economic dilution, but the correct net-share dilution is still roughly 16% at today's price before ordinary equity awards. The hedge protects the convertible-note leg; it does not make the warrants disappear.

Portfolio durability deserves the same accounting discipline. InterDigital capitalizes external filing and legal costs to obtain patents and amortizes internally generated patent costs over an estimated ten-year useful life. Acquired patents have averaged about 9.9 years of accounting life. Patent amortization was 70.7 million in 2025 against 54.6 million of new capitalized patent costs. That does not mean the patent estate is shrinking economically: legal patent terms and accounting lives differ, and portfolio count expanded sharply. It does show why maintenance investment belongs in owner earnings.

The count evidence is encouraging. The disclosed portfolio moved from roughly 27,500 patents/applications in 2021 to more than 30,000 by 2023, over 38,000 by the end of 2025 and more than 40,000 in current company materials. InterDigital is active in 5G-Advanced and early 6G work and describes substantial contributions to HEVC and VVC. Portfolio count is replenishing faster than expiry by simple unit count. The unresolved question is whether new patents carry comparable economic essentiality to the older cellular and HEVC assets. No public expiry ladder provides enough data to answer that quantitatively.

Business model, moat, industry and regulation

The company reports one GAAP operating segment because management allocates resources and evaluates performance at the consolidated level. Commercially, however, three licensing programs reveal the direction of travel.

Source: Q2 2026 10-Q.

The Streaming and Cloud number requires the most qualification. Amazon represented about 42% of Q2 company revenue according to the customer-concentration schedule, or roughly 109 million on a 260.2 million quarter. Streaming and Cloud revenue was 110 million. The near identity strongly indicates that almost all of the new program's Q2 revenue was Amazon. The company separately says the quarter's 103.7 million catch-up revenue was primarily related to Amazon. The exact Amazon recurring/catch-up split is not disclosed, so an estimate of about 5–6 million of current-quarter recurring Amazon revenue is an inference, not a reported number.

The split explains how a single streaming licence can transform headline mix without immediately producing 110 million of annual recurring streaming revenue. The commercial breakthrough is real; the quarterly segment mix exaggerates its recurring scale.

InterDigital's licence accounting has three broad economic forms. A static fixed-fee licence covering an existing portfolio can have performance substantially satisfied at inception and generate large point-in-time revenue. Dynamic fixed-fee arrangements give ongoing access to a changing portfolio and are recognized over the licence period. Running-royalty and hybrid agreements depend on usage or unit sales, with royalty revenue recognized as the relevant sales occur and estimates subsequently adjusted. Consideration attributable to prior unlicensed use can create catch-up revenue once a contract or other enforceable right allows recognition.

The structure creates extraordinary operating leverage. The research organization and patent portfolio have to be funded whether the next major licence signs in June or December. Legal teams may spend tens of millions before settlement. Once a licence signs, a large amount of historic consideration can pass through income with very little incremental quarterly expense.

Q2 operating costs show the other side. Research and portfolio development was 56.4 million, licensing expense 34.7 million and G&A 29.8 million, totaling 120.9 million. Total operating expense rose 27% year over year, including around 13.5 million of additional enforcement cost and 12.3 million of incremental stock-based compensation. For the first half, operating expense was 244.1 million, up 40%; the principal increases included enforcement, revenue-sharing obligations and compensation.

A sensible recurring cost base starts around 200–220 million of annual research and portfolio development, roughly 60–80 million of G&A and a licensing/legal line that can range far more widely. In a quiet year normalized operating costs might sit around the mid-300 millions. In an enforcement-heavy year they can exceed 400 million. Revenue can swing by far more than that, hence the apparently spectacular incremental margins on successful settlements.

Continuous R&D is mandatory. Research and portfolio-development expense was 195.3 million in 2023, 196.9 million in 2024 and 211.4 million in 2025, equal to roughly 36%, 23% and 25% of revenue respectively. Revenue is the wrong denominator when settlements create lumps, but the absolute expense trend shows InterDigital continuing to fund the research engine rather than harvesting an aging estate.

Three moat elements deserve genuine weight.

First, standards participation converts engineering work into legally enforceable claims before markets scale. An invention embedded in a cellular or video standard has a wider monetisation surface than a feature tied to one proprietary product. InterDigital's engineers continue to participate in 5G-Advanced, 6G and video standards work, and the company ranked among the top five U.S. applicants at the European Patent Office in 2025.

Second, portfolio breadth makes workaround increasingly expensive. A single patent can be invalidated or designed around. A portfolio spanning coding, HDR, streaming, wireless connectivity and device interaction can force an implementer to change several aspects of a service. Disney's experience is instructive: injunctions have affected HEVC encoding, HDR and casting-related functionality, and the company has experimented with technical workarounds instead of taking the demanded licence. InterDigital then expanded litigation against the alternative VP9 implementation. This does not prove that every claim will survive; it shows the economic logic behind portfolio breadth.

Third, legal endurance is a competitive asset. About 1.1 billion of liquid resources lets InterDigital run parallel proceedings in Europe, the United States, Brazil and elsewhere without needing a near-term settlement to fund operations. A poorly capitalized patent owner can be pressured by delay. InterDigital can carry the case.

A conventional customer-sticky moat deserves little weight. Licence counterparties frequently want to pay less and will litigate to do so. The durable advantage is enforceability, not satisfaction. Nor is brand a major economic moat. Regulatory licences, network effects and physical distribution are largely irrelevant.

Management has increasingly tailored the company around those realities. CEO Liren Chen took the role in 2021 after a career that included senior intellectual-property work at Qualcomm; Richard Brezski has been a long-tenured finance leader. That background matches the current licensing-enforcement strategy. The board has combined large buybacks with dividends and research investment. The 2023 repurchase program looks particularly successful ex post because much of the stock was retired far below today's price. By contrast, buying shares around 300 while a deeply in-the-money warrant structure approaches settlement produces less per-share accretion than the gross repurchase dollars suggest.

There is no dual-class founder structure or operating-company dependency that requires a governance discount comparable with controlled technology companies. Stock-based compensation does matter: its increase was a meaningful contributor to 2026 expense growth. I found no disclosed accounting restatement or auditor dispute that would make financial integrity a central thesis variable. PricewaterhouseCoopers remains the independent auditor in the 2025 filing.

The industry is better described as a standards-licensing ecosystem than a conventional "patent licensing market". The underlying economic base consists of billions of standards-compliant devices and, increasingly, services that use standardized video technology. Smartphone unit growth is mature, but each standards generation creates new technical claims and the opportunity to reset portfolio value. Video streaming broadens the royalty base from the device toward service providers. Connected cars and IoT create further endpoints.

InterDigital therefore experiences technology-iteration, policy and renewal cycles more strongly than ordinary macro cycles. Handset shipments matter most for running royalties. Fixed-fee arrangements mute short-term device demand. Interest rates affect valuation more than operating performance. The most dangerous "cycle" is the multi-year combination of patent expiry, standards transition and licence renewal: a portfolio that loses essentiality at the same time a major agreement expires can suffer a much larger reset than a temporary smartphone shipment recession.

The renewal calendar looks less threatening in 2026 than concentration alone suggests. Apple's current multi-year agreement runs through September 30, 2029 and was expected at inception to contribute approximately 133.7 million of annual revenue. Samsung's smartphone economics have also been locked in through a longer-term arrangement following arbitration. The 2025 10-K said fourteen agreements were due to expire by the end of 2026, but those agreements had contributed only about 15 million of 2025 recurring revenue. Five agreements that expired in 2025 and had not then been renewed represented about 31.7 million of recurring revenue; Xiaomi was subsequently renewed in January 2026 and Sony also renewed in early 2026.

Counterparty concentration remains severe despite that renewal protection. Amazon, Apple and Samsung each contributed at least 10% of Q2 revenue, together representing 68% of the quarter. The anonymized table assigns the three largest customers 42%, 13% and 13%. At June 30, three counterparties represented 83% of accounts receivable and contract assets. The precise mapping of anonymized customer letters to names is an inference from the disclosure order and should not be treated as explicitly confirmed.

The regulatory map currently leans somewhat more favourably toward patent holders in Europe and the United States than it did several years ago.

Europe's UPC can produce injunctions across multiple participating countries through one proceeding. That changes bargaining arithmetic: previously an implementer could force the patent owner to litigate country by country. Two InterDigital HEVC cases against Disney generated eleven-country injunctions; a third UPC ruling followed in September. Disney retains appeal rights, so the first-instance leverage has not yet become an irreversible appellate rule.

The European Commission's SEP Regulation proposal, which would have created new registration, transparency and conciliation mechanisms, was subsequently withdrawn. The removal of that proposal reduces one prospective source of centralized administrative friction for licensors, while existing EU competition-law constraints on seeking injunctions against willing licensees remain relevant.

The United States is moving rhetorically in a similar direction. The USPTO's December 2025 SEP Working Group announcement described an intent to restore robust and predictable remedies and explicitly defended injunctions and ITC exclusion remedies for valid patents. That is helpful policy language for InterDigital. U.S. district courts still apply the Supreme Court's traditional equitable injunction test, so policy support should not be translated into an automatic injunction entitlement.

China points the other way on one economically crucial question: who gets to set global FRAND rates. The EU filed a WTO dispute in January 2025 challenging China's assertion of power to set worldwide SEP licence terms. China's withdrawal of its anti-suit-injunction policy after a separate WTO ruling in April 2026 removes one escalation tool, but it does not settle the jurisdictional contest over worldwide rates. A sufficiently low Chinese global rate can weaken the value of European injunction leverage even when the European patents themselves survive.

The moat is strongest where technical breadth and multi-country enforcement reinforce each other; it is weakest wherever a court can convert that leverage into a global royalty materially below InterDigital's ask.

Horizontal competitor analysis

There is no truly direct listed twin. The relevant comparison set splits into three families: patent-licensing companies whose commercial relationship can be adversarial; technology licensors that transfer usable engineering IP; and hybrid businesses where royalties sit beside products.

Adeia is the nearest listed analogue to the first family. It licenses semiconductor and media intellectual property and, after the separation of Xperi's product businesses, derives economics primarily from IP licensing. Its similarity lies in monetising portfolios instead of shipping the end product. Its differences are equally important: InterDigital's identity is more tightly connected to cellular and video standards development and a large continuing research organization, while Adeia contains a larger element of acquired and legacy portfolio monetisation. Investors can use Adeia to ask what a patent-royalty cash flow is worth; they should not assume the portfolios have identical technical renewal rates.

Rambus sits one step further away. It remains an important memory-interface patent and technology licensor, but its current business includes physical memory-interface chips and other engineering IP. A customer can buy something from Rambus that is more product-like than an InterDigital patent licence. The stock is useful for seeing how markets value valuable semiconductor IP, but less useful as a pure royalty assertion multiple.

Dolby Laboratories has a large licensing business built around audio and imaging technologies. Dolby's commercial model has more brand, certification and collaborative technology adoption: manufacturers deliberately support Dolby formats because consumers and content creators recognize them. InterDigital usually monetises standardized functionality that implementers may prefer to take without entering a licence. Dolby deserves greater weight for royalty durability and less for litigation economics.

Universal Display owns foundational OLED intellectual property but also sells proprietary phosphorescent emitter materials. Its royalty margin is attractive, yet materials revenue exposes it to display production volumes and customer purchasing in a way InterDigital avoids. Q2 2026 Universal Display revenue fell to about 152 million as materials sales weakened, even while royalty revenue was more resilient.

Current market data make the valuation distinction visible.

†Peer market caps are market-data figures; InterDigital is calculated from the September 15 close and approximately 25.81 million basic shares because some vendors currently show a materially larger IDCC capitalization based on a diluted share count.

InterDigital is being valued much closer to Rambus's high-multiple IP growth story than to Dolby, Adeia or Universal Display. The premium can be rationalized only if the recurring base continues growing and UPC/streaming monetisation creates additional royalty categories. A mature patent portfolio whose revenue merely oscillated around 600–700 million would have difficulty supporting 38× trailing earnings indefinitely.

Arm is a useful negative control. Arm's fiscal 2026 revenue reached 4.92 billion, including 2.61 billion of royalty revenue and 2.31 billion of licensing revenue. Its customers pay for access to CPU and related technology products, tools, models and engineering ecosystems, then pay royalties as chips ship. Arm can retain royalties for decades because customers build product road maps directly around Arm's deliverables. The relationship produces a collaborative dependency and extremely long technology reuse. InterDigital has no analogous CPU-design hand-off. Applying Arm's multiple to InterDigital simply because both use the word "licensing" would ignore the difference that matters most.

Qualcomm supplies the opposite negative control. Its QTL licensing business is an economically important SEP royalty franchise, but QCT sells semiconductor products and gives Qualcomm a deep role in the same handset ecosystem in which it licenses patents. The licensing operation cannot be valued without acknowledging that semiconductor business, its customer relationships, and antitrust history. Qualcomm traded around 21× trailing earnings at the base date, well below InterDigital's quoted trailing multiple.

The patent pools are operationally more relevant than those two giant listed companies. Access Advance and Via LA aggregate portfolios from multiple patent owners and offer standardized licences. They can lower transaction costs and establish market reference points. A pool also gives up some bilateral pricing discretion and shares economics among members. InterDigital's direct enforcement model seeks the higher value that can come from negotiating its own portfolio, accepting the legal cost and volatility required to get there.

Customer choice is correspondingly different across the group. An Arm customer chooses Arm because using a proven architecture is cheaper and faster than building a CPU ecosystem from scratch. A Dolby customer adopts Dolby technology partly to sell a recognizable premium feature. A Universal Display customer needs emitter material and OLED patents. An Adeia counterparty often licences a portfolio to resolve or avoid IP exposure. An InterDigital counterparty signs because standardized wireless or video functions overlap patents that InterDigital can credibly enforce.

InterDigital occupies a niche between an R&D institution and a litigation-financed royalty collector. Its immediate profit pool comes from implementers of cellular and video standards. A rival research licensor can take that profit pool by contributing stronger patents to the next standards generation. A court can shrink it by setting lower FRAND rates. An open codec can erode a portion of video usage. A pool can commoditize licensing. No entrant can cheaply reproduce fifty years of patent history, but each new standards generation partially resets the contest.

The market's current comparison with Rambus is directionally understandable and numerically demanding. Rambus also carries a 38× trailing multiple, but it has a different product mix. Adeia's low-20s P/E gives a useful lower bound for what public markets currently pay for a more mature pure licensing stream. Dolby's mid-20s multiple is another relevant centre. On that evidence, InterDigital's premium reflects expected reacceleration rather than a generic IP-licensor valuation level.

Current fundamentals, accounting and litigation

The latest-four-quarter numbers are more informative in aggregate than as a smooth trend because individual quarters are settlement-driven. Q1 2025 had approximately 210.5 million revenue, derived from about 125.7 million recurring revenue and 84.8 million catch-up. Q2 2025 revenue was 300.6 million with 162.3 million of catch-up. Full-year 2025 revenue was 834.0 million. Replacing those first-half 2025 quarters with Q1 and Q2 2026 gives trailing-four-quarter revenue of roughly 788 million through June 2026. The figure remains substantial despite a different settlement calendar.

Q1 2026 revenue of 205.4 million beat the company's 194–200 million outlook. ARR reached 567.2 million, up 13% year over year, and smartphone ARR reached 491.8 million, up 18%. Management nevertheless left the then full-year revenue range at 675–775 million because the large Amazon development had not yet occurred.

Q2 changed the trajectory. Revenue of 260.2 million exceeded even the old quarter guide by more than 117 million at the top end. ARR increased again to 625.7 million, also 13% year over year. Full-year revenue guidance rose to 775–845 million; adjusted EBITDA was guided to 469–529 million, GAAP diluted EPS to 7.91–9.67, and non-GAAP EPS to 10.85–12.81. The raised guide is both a settlement story and a higher recurring-revenue story.

Q3 guidance of 154–158 million is revealing because its midpoint almost exactly matches the 156.4 million of Q2 recurring revenue. The match suggests management is not relying on another Amazon-sized catch-up event to establish the near-term quarterly floor. A Q3 result near guidance with ARR still around or above 625 million would be qualitatively better than another 250 million quarter dominated by a large historical lump.

The market is trading three themes at once.

The first is recurring licensing growth. ARR has moved from 502.9 million in Q1 2025 to 567.2 million in Q1 2026 and 625.7 million in Q2. That is genuine improvement.

The second is a new streaming royalty pool. Amazon's agreement proves that a major streaming platform can become a direct InterDigital licensee rather than relying solely on device manufacturers to carry video royalty economics. Disney is being litigated precisely over the same broad idea. The addressable revenue can expand meaningfully if the Amazon outcome becomes a repeatable commercial benchmark.

The third is injunction leverage. The first two UPC injunctions cover eleven EU countries; the September ruling added another UPC victory. National German and Brazilian injunctions broaden the pressure. Disney has nevertheless chosen workarounds and appeals rather than immediate capitulation, which is useful bear evidence: an injunction does not automatically convert into the licensor's requested royalty.

The Amazon accounting warrants a closer walk-through.

InterDigital's 2025 10-K says revenue contracts are accounted for under ASC 606's five-step model. Where consideration is uncertain, the company estimates transaction price only when a contract exists and includes variable consideration only to the extent it is probable that a subsequent change will not result in a significant reversal of cumulative revenue. Fixed-fee agreements can be recognized at inception or over time depending on whether the obligation is static access to existing IP or dynamic access to an evolving portfolio.

Amazon created an enforceable licence in Q2 even though price was unresolved. That distinction allows revenue recognition to begin. The 10-Q explicitly says the company began recognizing revenue using a conservative GAAP estimate pending binding arbitration. The estimate is included in the schedule of fixed-fee contracted revenue, together with the estimated Lenovo consideration.

The future contracted fixed-fee revenue schedule was approximately:

The company expressly states that this schedule includes estimates for arrangements whose final amounts are being arbitrated, limited by ASC 606's significant-reversal constraint. It should be viewed as contracted-plus-constrained-estimate revenue, not a cash-guaranteed order backlog.

At June 30, current and non-current contract assets together had risen to roughly 175 million from around 41 million at year-end 2025. A contract asset means revenue has been recognized before an unconditional billing right exists. Amazon is an obvious contributor to the changed accounting environment, but the company does not provide an Amazon-only contract-asset balance, so assigning the increase entirely to Amazon would overstate the disclosure.

The central Amazon fact is that recognised revenue is an ASC 606-constrained estimate, not the arbitrator's rate. The company has disclosed neither the estimate itself nor a dollar sensitivity showing what a lower award would reverse.

That asymmetry should affect valuation. An eventual award above the estimate generates upside. An award below it can reduce previously expected economics and require the transaction-price estimate to be revised. The accounting constraint lowers reversal risk; it does not turn a pending arbitration into a fixed receivable.

The litigation docket reinforces why this is normal business activity.

Amazon's underlying disputes across various jurisdictions were resolved, stayed, withdrawn or terminated following the licence. The economic dispute migrated from infringement litigation to binding arbitration. InterDigital has described the expected arbitral process as a roughly 18–24 month matter, so final economics are more naturally a late-2027/2028 catalyst than a near-term quarterly event.

Disney is the largest live enforcement campaign. The June Mannheim UPC judgment upheld an HEVC patent and granted an eleven-country injunction; Düsseldorf did the same with a separate HEVC patent in July. The September 2 Düsseldorf decision added a third UPC injunction involving cross-device video functionality, while German national courts and Brazil had already produced other injunctions. The company has continued to seek damages and challenge Disney workarounds. Appeals remain available.

Transsion remains another important example because a Brazilian court found InterDigital's offer FRAND and awarded an injunction, according to the company's Q1 disclosure. That result gives InterDigital another jurisdictional precedent outside Europe, although Brazil has a much shorter SEP case history than the UK or Germany.

Automotive licensing carries a different legal risk. In July 2026 the UK Supreme Court allowed Tesla to revive its effort to have an English court determine global FRAND licensing terms involving InterDigital and the Avanci pool. The substantive rate has not yet been determined, but the jurisdictional result matters because it can provide implementers with another route to global court-set terms.

Lenovo supplies the best public base-rate evidence. The first UK global FRAND determination produced roughly 0.175 per unit, and the Court of Appeal subsequently raised it to around 0.225. InterDigital's trial position was materially higher, around 0.498 per unit. Exact comparisons between an opening offer and a final judicial rate depend on volumes, generations, time periods and payment timing, but the central observation survives those adjustments: the public court-set outcome was well below InterDigital's preferred economics.

Samsung points in the opposite direction. Its arbitration produced an agreement worth roughly 131 million of recurring annual revenue, about 67% above the economics of the prior Samsung licence according to company discussion. That shows InterDigital can achieve meaningful rate step-ups. The confidential opening positions prevent calculation of an award-to-demand ratio.

The statistically honest base-rate conclusion is modest: there are too few observable cases. Arbitration keeps the most important negotiating inputs private. The one well-documented court-set global FRAND rate, Lenovo, imposed a substantial haircut to the licensor's position. Samsung shows that a private tribunal can still deliver a large economic increase. Anyone modeling all future disputes at InterDigital's requested rate is using a more bullish assumption than the public evidence supports.

Concentration compounds this uncertainty. Q2's three ≥10% customers accounted for 68% of revenue. A single non-renewal or adverse rate reset can create a visible step-down even while the overall patent estate remains healthy. The mitigating fact is that Apple runs through 2029, Samsung is on a long-duration arrangement, and the agreements expiring through 2026 represent a relatively small disclosed recurring-revenue pool.

The full unlicensed pipeline is not publicly enumerated. Disney is a confirmed major unlicensed/disputed streaming counterparty. Amazon has moved from unlicensed litigant to licensee. Calling every other major streamer an "unlicensed target" would go beyond the filings. That distinction matters when constructing a streaming TAM: an addressable service market is not the same thing as a proven patent royalty pool.

Valuation, risks, catalysts and cross-synthesis

The current close is 326.08 as of Tuesday, September 15, 2026. Using approximately 25.81 million basic shares gives a basic equity value of about 8.42 billion. The six-million warrants add about 4.06 million economically dilutive shares at the present price, giving 29.87 million before employee awards and about 9.74 billion of warrant-adjusted equity value. Q2 non-GAAP EPS arithmetic points to roughly 30.6 million diluted weighted shares after the ordinary equity instruments are included; at the same price that is close to 9.98 billion.

June net liquidity, before considering future note settlement mechanics, was roughly 720 million after subtracting the approximately 380 million convertible principal from about 1.1 billion of cash and short-term investments. Paying the convertible principal in cash reduces both cash and debt and does not destroy that net position mechanically. The purchased hedge offsets much of the conversion spread; the sold warrants remain the principal economic dilution.

At 326.08, the quoted trailing P/E from market data is about 38×. On the company's 2026 non-GAAP EPS guidance of 10.85–12.81, the price is roughly 25.5–30.1×, with a midpoint around 27.6×. On GAAP guidance of 7.91–9.67, it is about 33.7–41.2×. Those are demanding multiples for a business whose recurring base is growing around the low teens and whose legal outcomes remain binary enough to move annual revenue by nine figures.

The historical percentile is less precise than it looks on screening services. A P/E series divides a volatile settlement-driven numerator into price, so a quarter with a 100 million catch-up can make the stock appear suddenly cheap without changing the portfolio's long-run value. I would describe the present multiple as upper-range rather than assign a false exact percentile. The more reliable cross-check is that InterDigital currently trades near Rambus's approximately 38× P/E and materially above Dolby, Adeia, Universal Display and Qualcomm.

The ARR multiple is starker. Taking roughly 30.6 million diluted shares at 326.08 gives nearly 10.0 billion of equity value. Subtract around 0.72 billion of net liquidity and economic enterprise value is about 9.3 billion. Against Q2 ARR of 625.7 million, that is roughly 14.8× EV/ARR. ARR is revenue, not gross profit, but this is still a large multiple for an accounting run rate that includes estimated licence economics and requires around 200 million a year of research to regenerate the asset.

The long-term growth expectation embedded in the bull case is understandable. Moving from 625.7 million ARR in Q2 2026 to 1 billion around 2030 requires roughly 11% annualized growth. That is close to the recent year-over-year ARR growth rate. The market is capitalising a continuation of current momentum rather than an arithmetically impossible target. The hard part is the multiple investors will pay once the target has already been reached.

Cash-flow passthrough. The 2023–25 operating-cash-flow/net-income ratio was about 1.05×. Over the broader five-year filing record the direction remains cash-generative rather than cash-consuming, but annual timing is extreme. For owner earnings I deduct all property/equipment spending and capitalized patent costs, even though a portion of patent additions is growth investment. This is deliberately conservative because a patent licensor cannot stop replenishing the estate and still be valued as a going concern.

On that basis, 2023–25 owner FCF was about 169 million, 213 million and 474 million. The three-year average is roughly 285 million. Against the roughly 9.98 billion diluted equity value, normalized three-year owner-earnings yield is only about 2.9%. Using exceptional 2025 FCF alone raises the yield to about 4.7–4.9%; using H1 2026 annualized cash flow lowers it sharply. The normalized figure is more defensible. The gap between normalized owner earnings and headline earnings is below the 30% threshold that would force the owner-earnings basis, so accounting earnings need not be discarded, but cash-flow normalization materially changes how attractive a single year looks.

I use three valuation anchors: normalized owner earnings, forward earnings and recurring-revenue economics. An explicit DCF is useful as a cross-check but can give false confidence because the timing of legal receipts is unknowable. All scenarios below use roughly 30–31 million economic diluted shares, in place of the 25.8 million basic count.

†Approximate total-return CAGR includes around 9 of cumulative dividends over three years and assumes no material change to dividend policy. These are valuation scenarios within a research framework, not investment advice.

The conservative case treats recent recurring growth as partly cyclical to unusually productive licence signings. ARR reaches only 750–800 million by 2030, owner earnings normalize around 300 million, and a mature patent licensor receives an 18–20× multiple. That produces approximately 200–230 per share.

The base case assumes InterDigital gets close to the billion-dollar recurring target, smartphone economics remain stable, video/CE continues expanding, and at least some streaming counterparties follow Amazon. Normalized owner earnings move toward 400 million or slightly above. A 20–22× multiple is still a premium to mature licensing comparables but recognizes portfolio growth, cash generation and balance-sheet strength. The result is approximately 285–340 per share.

The optimistic case requires a real second growth curve: Amazon establishes attractive streaming economics, Disney eventually licences at a commercially meaningful rate, additional services follow, and 6G/VVC work replenishes the patent estate. ARR reaches 1.15–1.25 billion and normalized owner cash generation exceeds 500 million. A 22–24× owner-earnings multiple supports roughly 400–475.

These assumptions expose the expectation gap. At 326.08, investors have already paid for much of the base case. A further material re-rating needs evidence that the optimistic scenario is becoming normal rather than exceptional. The market will care more about recurring ARR and new program licences than about another large historical catch-up quarter.

The most fragile base assumption is that recurring revenue can reach around one billion without a major legal/policy reset. Cutting the incremental growth embedded in that assumption to roughly 70% moves 2030 ARR closer to the mid-800 millions and normalized owner earnings toward 340–360 million. At a 20× multiple plus the balance sheet, that reduces my base value toward roughly 250–275 per share.

The margin-of-safety test gives a clear answer. Current price is about 42–63% above the 200–230 conservative value range. Margin of safety against that case is therefore zero.

The flat-earnings test is equally demanding. With annual dividends of 3.00, a flat share price and no earnings growth produce only about a 0.9% annual cash yield. U.S. ten-year Treasury yields were around the 4% area during 2026; the Treasury's official series, for example, showed 4.44% on March 27. I was unable to retrieve a reliable September 15 daily Treasury row before the research cut, so I do not fabricate a base-date decimal. The comparison does not depend on that missing decimal: 0.9% is far below prevailing long-term government yields. There is no margin of safety at this buy price.

【Margin-of-safety sufficiency verdict】none

That does not make the underlying company poor. It makes entry price a material part of the thesis.

The permanent-loss risks concentrate in five places.

The first is rate-setting risk, with medium probability and high impact. Lenovo is the observable warning. If Amazon's arbitral economics or future global FRAND determinations land materially below InterDigital's internal estimates and negotiating positions, recurring revenue, prior constrained estimates and the multiple can all be hit together. The indicators are court/arbitration awards, changes in contract assets and any revenue-estimate revisions.

The second is UPC reversal or erosion, with medium probability and high impact. The current Disney decisions give InterDigital unprecedented multi-country leverage, but they remain subject to appeal. A material Court of Appeal reversal on validity, infringement, FRAND conduct or injunction standards would damage more than the specific Disney royalty; it would weaken the market narrative that Europe has permanently shifted negotiating power toward SEP holders. The observable indicators are stays, appellate judgments and whether other UPC divisions follow or distinguish the Disney rulings.

The third is portfolio-replenishment risk, low-to-medium probability over twelve months but high over five years. Patent count is rising, which is encouraging; economic quality is harder to observe. A 6G or post-HEVC world in which InterDigital owns fewer genuinely essential claims could leave the company funding a 200-million-plus research engine while older high-value patents expire. Watch standards contribution rankings, new patent families, independent essentiality studies and licence coverage of next-generation standards.

The fourth is concentration, medium probability and medium-to-high impact. Three Q2 counterparties produced 68% of revenue and three counterparties represented 83% of receivables and contract assets. Long-duration Apple and Samsung arrangements reduce immediate renewal danger; the concentration means any adverse interpretation, credit dispute or accounting change is still disproportionately visible.

The fifth is valuation and dilution, high probability of affecting returns and medium impact absent a fundamental failure. The stock is already valued at roughly 38× trailing earnings and nearly 15× diluted EV/ARR, while around four million warrant-equivalent net shares are economically in the money. Even perfect operational execution can produce mediocre equity returns if the terminal multiple moves from the high-20s/30s toward the low-20s.

Positive catalysts over the next twelve months are more concrete than the Amazon arbitration itself. They include ARR holding above the 625.7 million Q2 level, another large streaming or video licence, a Disney settlement following the accumulating injunction pressure, additional favourable UPC appellate or first-instance rulings, and cash conversion improving after the enforcement-heavy first half. Continuing buybacks would be positive only if the price offers better economics than the shares potentially being created by warrants.

Negative catalysts include Disney obtaining an appellate stay or substantive reversal, a court accepting a materially lower global FRAND framework, a downward Amazon revenue-estimate adjustment, Q3/Q4 ARR dropping despite a superficially met annual revenue guide, and enforcement costs remaining above the new recurring revenue they produce.

The practical tracking dashboard is:

The financial indicators come from the 10-Q and earnings release. The October 29 earnings date is an estimate based on InterDigital's established late-October reporting cadence; as of the research cut I did not find a company announcement fixing the Q3 date, so it should not be treated as a confirmed calendar event.

ARR and recurring revenue belong at the top because they distinguish durable licensing growth from settlement timing. Contract assets are the best visible accounting warning light for revenue recognized ahead of an unconditional right to bill. Research spending is the long-horizon portfolio-health gauge. Customer concentration measures how much one negotiation can still move the company. Warrant dilution should be recalculated with every material share-price move because net settlement increases as the stock moves further above 105.43.

The cross-synthesis is now relatively clear.

Vertically, InterDigital has proven one capability for more than five decades: it can keep repositioning communications research into assets somebody else eventually pays to use. The company survived a product-company origin, unsuccessful ambitions in semiconductor commercialization, the rise of 2G/3G/4G/5G, years of licensing disputes and the maturation of smartphones because management kept the research institution while changing the commercialization layer. The 2012 Intel patent sale validated the financial value of the portfolio. The Technicolor-related video expansion broadened it, and the recent smartphone and streaming agreements are testing whether that portfolio can become a larger, less handset-dependent annuity.

That historical success was partly technological and partly institutional. Cellular standardization created an enormous tailwind. A company with useful communications inventions could reach products built by Samsung, Apple, Xiaomi and others without manufacturing any of them. InterDigital also made an important strategic choice: it stopped trying to win the product race and concentrated capital around research plus licensing. Litigation expertise and a large liquid balance sheet then became complements to technical IP.

Those success factors still exist. Portfolio count is expanding, research spending remains above 200 million annually, 6G and VVC work is active, and the licensing base has moved beyond smartphones. The market has good reason to assign more value to the company than it did when each major handset renewal could dominate the entire story.

Horizontally, the real advantage is independence. Arm must build engineering products customers actively want. Qualcomm must compete in chips while also licensing. Universal Display relies partly on materials. Rambus sells physical interface products beside licensing. InterDigital can direct nearly all its commercial capital toward research, portfolio management, negotiation and enforcement. The weakness is the mirror image: it has fewer ordinary commercial relationships to cushion an adverse patent ruling. A court can attack the monetisation mechanism directly.

The UPC currently improves that equation. The Disney cases turn fragmented European enforcement into a credible eleven-country threat. The Amazon licence shows that streaming services may prefer negotiated access to service degradation and parallel litigation. A few more outcomes of that kind would make streaming a genuine second royalty pool.

The market may nevertheless be overestimating how quickly legal leverage becomes permanent cash economics. Disney has absorbed injunctions and changed technical features rather than immediately signing. Amazon's price is unresolved. The visible Lenovo court-set rate was much lower than InterDigital wanted. Tesla has just secured another opportunity to seek a court-set global FRAND framework in the UK. Those facts make the current valuation qualitatively different from paying 20× a settled annuity.

For the next year, ARR matters most. A revenue beat driven by another hundred million of catch-up would be less informative than recurring revenue moving from 625.7 million toward 675–700 million. Disney is the next major commercial proof point. The warrant structure also enters its settlement window around September 2027, so dilution will become increasingly visible.

For the next three years, Amazon's arbitral economics, additional streaming licences and the durability of UPC doctrine matter most. This is the period when the market learns whether the new streaming category is a repeatable business line or simply a handful of high-profile disputes.

For five years, portfolio renewal dominates everything. A patent licensor can look wonderful while harvesting a generation of SEPs and deteriorate abruptly when those rights age. The count trajectory, 6G contribution quality, VVC adoption and licensing coverage of post-5G standards will decide whether InterDigital deserves to be viewed as a compounding research institution or a declining royalty estate.

The stock becomes substantially more attractive under either of two conditions: valuation falls far enough that the existing recurring base alone supports the price, or fundamental evidence moves far enough that 1 billion-plus ARR no longer requires optimistic assumptions. My preferred version is the first because it does not require guessing court outcomes.

【Bull reasons】

• Q2 ARR reached 625.7 million, up 13% year over year, while smartphone ARR had already risen 18% year over year in Q1, giving the growth thesis a recurring component rather than relying solely on settlements.
• Amazon establishes a direct streaming-and-cloud licensing category, while three UPC injunctions against Disney give InterDigital tangible bargaining leverage with another major service provider.
• The portfolio expanded to more than 38,000 patents/applications by year-end 2025 and more than 40,000 in current materials while research and portfolio expense exceeded 200 million, supporting replenishment by count.
• Around 1.1 billion of liquidity allows the company to litigate for years without settling under financing pressure.
• Near-term renewal risk is smaller than concentration implies: Apple runs to 2029 and the agreements scheduled to expire by end-2026 accounted for only about 15 million of 2025 recurring revenue.

【Bear reasons】

• At 326.08 the stock trades around 38× trailing earnings and roughly 15× diluted EV/ARR, already capitalising a substantial portion of the billion-dollar ARR thesis.
• The public Lenovo rate-setting precedent produced a rate materially below InterDigital's litigation position, showing that legal victory on essentiality does not guarantee the licensor's requested economics.
• Amazon's final price is unresolved, yet revenue is already being recognized using a constrained estimate; the company gives no Amazon-specific downside sensitivity.
• Six million deeply in-the-money warrants translate to roughly 4.06 million net incremental shares at the current price, before normal employee-equity dilution.
• Three counterparties generated 68% of Q2 revenue and three counterparties represented 83% of receivables and contract assets, leaving the economics highly sensitive to individual negotiations.

【Pre-mortem】

One plausible 50%-loss script runs through 2027–28. Disney wins a material UPC appeal or establishes workable codec alternatives, limiting the commercial value of InterDigital's injunctions. Amazon's arbitration lands below the constrained estimate or provides only modest recurring economics. ARR stalls around 650–700 million instead of approaching 800 million. Normalized earnings settle around 8–9 per diluted share and the market stops paying a growth multiple, assigning 18–20× instead. A price around 145–180 would be roughly 45–55% below the September 2026 close.

A second script is more structural. UK, Chinese and other courts increasingly set global FRAND rates closer to Lenovo-style judicial outcomes while UPC appellate doctrine makes injunctions harder against implementers willing to submit to court-set terms. By 2029 the older high-value wireless/video estate is rolling off faster in economic quality than 6G/VVC patents are monetized. Research and legal costs remain near 400 million, normalized EPS falls toward 7, and an 18× multiple produces roughly 125–130. That path would represent a permanent impairment rather than a temporary share-price drawdown.

【Research uncertainties】

The largest blind spot is Amazon's undisclosed constrained transaction-price estimate. Neither the provisional total consideration nor a downside sensitivity is public.

The second is private-arbitration data. The opening positions in Samsung and other arbitrations are confidential, making a statistically sound award-versus-demand base-rate study impossible. The public Lenovo case is informative but only one major observation.

The third is patent maturity. InterDigital gives portfolio counts and accounting amortization but no essentiality-weighted expiry ladder. Count growth cannot prove economic replenishment.

The fourth is the unlicensed-counterparty list. Disney is visible because of litigation; the complete universe of streaming services, device makers and manufacturers that management believes require a licence is not publicly disclosed.

The fifth is UPC appellate durability. Three Disney wins are meaningful first-instance evidence, but the most important structural conclusions still await more appellate jurisprudence.

【Primary source hierarchy】

The analysis gives greatest weight to InterDigital's Q2 2026 Form 10-Q and earnings release, the 2025 Form 10-K, the 2026 proxy, SEC filings governing the convertible hedge/warrants, European Commission and USPTO policy material, and company/court disclosures around litigation. Secondary IP-law reporting is used where primary court material was not readily machine-accessible, particularly for the September Disney decision and historical FRAND-rate comparisons. Market prices are checked against September 15, 2026 trading data.

The final judgment follows from the interaction of quality and price.

InterDigital has become a better company than its old "lumpy patent lawsuit" label implies. Recurring revenue is higher, the portfolio is expanding, smartphone licences are better covered, video research has opened additional royalty categories, and the UPC gives the patent estate a stronger enforcement channel. The balance sheet can finance the strategy without depending on capital markets. I regard fundamental quality as above average for an IP monetisation company.

The present equity price already asks investors to believe a large part of that improvement will persist. A roughly 326 share price equates to close to 10 billion of economic diluted equity value, a high-20s multiple of 2026 midpoint non-GAAP EPS and roughly 15× diluted EV/ARR. Amazon terms are still being arbitrated; Disney remains unlicensed and is appealing/fighting; public court-set FRAND evidence is mixed; and roughly four million warrant-equivalent net shares sit above the basic count. The company can execute well and still deliver modest returns if valuation normalizes.

For an existing holder with a low basis, the recurring growth and optionality justify retaining exposure. For fresh capital, I think waiting produces better expected asymmetry. A price in the high 100s would allow the present portfolio rather than future litigation wins to carry more of the valuation. Alternatively, materially higher ARR with finalized Amazon/Disney economics could raise the conservative value and justify entry at a higher market price.

【Company-profile scores】

• Fundamental quality: high
• Growth: medium
• Moat: strong
• Financial soundness: strong
• Management credibility: high
• Valuation attractiveness: low
• Risk level: high
• Suitable investor type: event-driven / long-term value investors comfortable with patent litigation and accounting complexity

【Investment rating】

• Rating: Hold
• One-line thesis: Rising recurring royalties and stronger UPC enforcement are real, but 326.08 already prices substantial streaming success while Amazon economics and dilution remain unresolved.
• Current-price classification: acceptable hold
• Whether to wait for a better price: yes. For new money, I would require either the 165–195 price area or a fundamental reset in which ARR and finalized streaming economics raise the conservative value materially.
• Target holding horizon: 3–5 years
• Expected annualized return: conservative about -12%; base around 0%; optimistic about +11% over three years, including approximate dividends.
• Max-loss risk: roughly 50–60% in a combined adverse-rate, UPC-reversal and multiple-compression case, with a plausible stressed range around 125–180.
• Reassessment-trigger signals: ARR below 590 million for two consecutive quarters; an Amazon transaction-price downward revision; a substantive UPC appellate reversal on Disney FRAND/injunction issues; normalized annual research spending below about 180 million; or economic net warrant dilution plus equity awards rising materially beyond the roughly 30–31 million share valuation base.

【Ideal Buy Price】165–195 USD

Basis: at least about a 20% discount to the 200–230 conservative intrinsic-value scenario, with extra allowance for arbitration uncertainty and warrant dilution.

Acceptable hold price: 265–355 USD, approximately the ±15% neighbourhood of the base-case valuation centre.

Clearly overvalued price: 525 USD and above, approximately 10% or more above the upper end of the optimistic intrinsic-value scenario.

【Valuation Range】

• current: 326.08 (close as of 2026-09-15)
• bear (conservative · ideal buy zone): [165, 195]
• base (fair · acceptable hold zone): [265, 355]
• bull (optimistic · above the clearly-overvalued line): [525, 575]

Other tickers mentioned

• RMBS.US: Rambus is the high-multiple listed IP comparison, but its memory-interface product revenue makes it less pure than InterDigital.
• DLB.US: Dolby provides a mature licensing comparison with a more collaborative branded-technology model.
• ADEA.US: Adeia is the nearest listed pure patent-monetisation comparison among the principal peers considered.
• OLED.US: Universal Display combines foundational OLED licensing with physical emitter-material sales, limiting direct comparability.
• ARM.US: Arm illustrates the engineering-IP model in which customers receive CPU technology products and subsequently pay per-chip royalties.
• QCOM.US: Qualcomm illustrates the hybrid model combining a major wireless patent-licensing operation with a semiconductor business.
• AMZN.US: Amazon is InterDigital's first disclosed major Streaming and Cloud licensee, with final economics pending arbitration.
• AAPL.US: Apple is a major recurring smartphone licensee under an agreement running through September 2029.
• TSLA.US: Tesla's UK action against InterDigital and Avanci tests another route to court-set global automotive FRAND terms.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.