Medical AI 4S: Claude wins another victory! The field continues to heat up, which giants are entering the game?

Just now, Anthropic unveiled its first achievement from its new molecular biology lab: the large model Claude has independently discovered a completely new enzyme system in bacteriophage DNA that has never been recorded by humans before!

Its structure is remarkably similar to CRISPR, the "scalpel of God," possessing the ability to cut, copy, and paste DNA. The entire discovery process utilized 950 AI agents, ran for 21 hours, and consumed approximately 210 million tokens. Subsequently, human scientists took over the review and experimental verification.It can be said that from today onwards, the research paradigm of life sciences has been completely rewritten.The milestone of accelerating medical AI4S is approaching.

I. What happened? Claude enters the AI medicine field.

When general-purpose models like Claude and ChatGPT first entered the real-world R&D procurement lists of Pfizer, Novo Nordisk, and Eli Lilly, the scarce asset of AI-driven drug development shifted from "model capabilities" to a closed loop consisting of "model + proprietary data + wet lab feedback." This is the core logic behind why giants like Anthropic are directly involved in drug development.

In addition to technological breakthroughs by industry giants, the AI pharmaceutical sector has received clear boosts in terms of policy, industry, and funding from a macro perspective.

Policy-wise: AI-driven drug development is included in the 15th Five-Year Plan.

On September 18, ten departments, including the Ministry of Industry and Information Technology, jointly issued the "15th Five-Year Plan for the Development of the Pharmaceutical Industry." Three points in the plan are directly related to AI-driven drug development: by 2030, first-in-class (FIC) drugs will account for more than 25% of the global market; the average annual growth rate of the innovative drug industry will reach more than 20%; and the average annual R&D investment intensity of listed pharmaceutical companies will reach more than 10%. Among these, "Cultivating High-Value Application Scenarios for AI-Driven Drug Development" is listed as a separate section, explicitly identifying AI-enabled target screening, drug molecule design, and modification as key high-value application scenarios requiring cultivation. At the level of key tasks, the plan proposes to "accelerate the application of new technologies such as artificial intelligence, quantum computing, supercomputing, and computational medicine to empower drug development," and further requires "accelerating the paradigm shift in pharmaceutical R&D and production management driven by artificial intelligence."

This plan elevates AI-driven drug development from a "technical means" to an "industry direction" for the first time, placing it within the same indicator system as the two most crucial output metrics: global FIC share and innovative drug growth rate. This will directly alter the behavioral expectations of two key stakeholders: for innovative pharmaceutical companies, AI will transform from a "cost reduction tool" into a "means to achieve FIC share targets"; for upstream life science tool and service providers, the R&D output released by AI will ultimately translate into orders.

If the goals for innovative drugs are achieved, the annual scale of China's innovative drug industry by 2030 will be nearly 2.5 times that of 2025, with the newly added industry scale exceeding 1.5 times the current size within five years. This growth rate is rare in the historical context of the global pharmaceutical industry: the compound annual growth rate of the global prescription drug market over the past decade has been roughly in the range of 5%-7%, and even if innovative drugs are considered separately, there are few cases in major markets that have maintained 20% growth for five consecutive years. This also explains why the planned AI column is not a dispensable appendix: under the traditional R&D paradigm, relying on manpower and trial and error to achieve such growth would lead to rapidly increasing marginal costs; including AI in the column essentially means that the path to achieving the growth target is partly based on a technological leap in R&D efficiency.

Industry side: Nine events in nine days, four of which occurred on September 16th alone.

If we view policy as a top-down impetus, then the nine events from September 10th to 18th represent bottom-up industry evidence. The sheer density of these events constitutes information in themselves—within nine days, an AI-native company's core pipeline entered Phase III, two US AI and computing power giants signed cooperation agreements with multinational pharmaceutical companies (MNCs), three MNC open platforms simultaneously connected to external data providers, a Chinese innovative pharmaceutical company completed a $750 million overseas transaction for a trispecific antibody, and finally, a five-year plan was released.

Event 1: The world's first AI-driven end-to-end drug enters Phase III. On September 10th, Insilicon's Rentosertib (a TNIK inhibitor indicated for idiopathic pulmonary fibrosis) completed its first patient dosing in the GENESIS-IPF-3 Phase III clinical trial in China. What makes this drug unique is that both target discovery and molecular design are completed by an end-to-end AI platform, rather than algorithms being partially embedded in traditional R&D processes. Previous Phase IIa data showed that after 12 weeks of once-daily administration of 60mg, patients' forced vital capacity increased by an average of 98.4 ml, while the placebo group experienced a decrease of 20.3 ml. The significance of this data lies in the fact that it advances "AI-driven drug development" from the PowerPoint stage to a stage where clinical endpoints can be used for measurement.

Event Two: Large-Scale Models Officially Enter MNCs' R&D Procurement Lists. On September 16th, Novo Nordisk announced a collaboration with Anthropic to test Anthropic's Claude Science research platform on drug discovery problems selected by Novo Nordisk scientists and computing teams, and to leverage its cutting-edge models to enhance AI-driven software development. Novo Nordisk President and CEO Mike Doustdar's statement is noteworthy—he said this collaboration is "a further extension of Novo Nordisk's current AI initiatives with other technology partners," and explicitly stated that the purpose of AI is to "improve R&D productivity and shorten the path from research to market." Anthropic co-founder and CEO Dario Amodei's statement offers a grander vision: "As AI capabilities continue to improve, it is expected to compress a century's worth of biological and medical breakthroughs into a decade." This is not just a software purchase order, but rather the integration of a general-purpose large-scale model research platform into the R&D processes of multinational pharmaceutical companies—meaning that large-scale model vendors have, for the first time, gained access to pharmaceutical R&D budgets, rather than simply selling computing power or APIs to pharmaceutical companies.

Events 3-5: The MNC Open Platform integrated three external data providers on the same day. Also on September 16th, Eli Lilly's TuneLab platform (a collaborative AI/machine learning drug discovery platform launched by Eli Lilly in September 2025) secured two collaborations: Twist Bioscience became its preferred provider of antibody characterization data services, generating high-throughput wet experimental data for AbLab (Lilly's antibody druggability prediction model); Genscript Biotech joined on the same day, providing wet experimental services for protein expression, purification, and characterization to platform member companies, transforming AI predictions into biological validation data. Combined with Schrödinger, which had previously integrated the TuneLab workflow into its LiveDesign enterprise information platform, Eli Lilly's TuneLab now forms a three-tiered structure in the antibody field: "model provider + platform provider + data provider." Furthermore, on September 16th, Novartis exercised its exclusive option to acquire global rights to Sironax's proprietary brain delivery platform, with Sironax receiving $125 million upon completion of the transaction.

Events 6-9: Chinese innovative pharmaceutical companies' overseas expansion efforts are simultaneously taking shape. On September 15th, GSK announced an agreement with Enmu Biotechnology to acquire global rights to a trispecific T-cell engager (TCE) antibody for multiple myeloma, with a maximum transaction value of $750 million. On the same day, Insil Biotech announced its "longevity vaccine" research plan, focusing on guiding patients' own immune cells to clear the earliest cell-driving factors associated with aging-related diseases. On September 18th, Baili Tianheng announced that its wholly-owned subsidiary, Pangu Capital, plans to subscribe for 3,910,952 Series A preferred shares of Connexus Therapeutics for $5 million. Based on the exercise of all reserved employee incentive shares, the shareholding ratio after the capital increase will be 7.91%.

Connecting these events reveals a clear causal chain: large model manufacturers need real-world scientific questions to demonstrate the research capabilities of their models; microcomputer manufacturers (MNCs) need external AI capabilities to offset the pressure on their self-developed pipelines; AI-native companies need MNC orders and clinical resources to complete closed-loop validation; and all these stages ultimately require large-scale, standardized wet laboratory data for practical application. The needs of these four entities converge at the same point—data closure.

From a funding perspective: the leading gainers weren't algorithm companies, but rather those who sold shovels for wet experiments.

Policy and industry events will ultimately leave their mark on prices. In the week of September 14-18, 2026, both A-shares and Hong Kong stocks in the pharmaceutical sector rose: the Shenwan Pharmaceutical and Biological Index rose 3.26%, outperforming the CSI 300 Index by 3.32 percentage points, ranking 4th among the 31 Shenwan first-level industries; in Hong Kong, the Hang Seng Index fell 0.22%, while the Hang Seng Healthcare Index rose 4.07%, outperforming the Hang Seng Index by 4.30 percentage points; the Wind Hong Kong Healthcare Index rose 4.10%, ranking 1st among the 11 Wind industries.

Structure is more informative than total volume. All 13 sub-sectors of the A-share Shenwan Medical and Biological sector closed higher this week, but the gains were extremely unevenly distributed: Medical R&D outsourcing (CRO/CDMO outsourcing services within CXO) rose 8.32%, other biological products rose 5.40%, and APIs rose 5.17%, forming the top tier; while hospitals rose only 0.18%, traditional Chinese medicine III rose 0.30%, and blood products rose 0.89%, ranking at the bottom. The difference between the top and bottom performers was 8.14 percentage points. This pattern of "all rising but extremely divergent" indicates that the funds buying into the pharmaceutical sector this week were not making sector allocations, but rather choosing a direction—the result of buyers voting with their feet is that outsourcing and biological products, directly related to AI R&D, have a significantly higher priority than end-user medical services and traditional Chinese medicine.

Comparing the direction of capital flow with the facts reveals a self-consistent chain: policies set targets for the proportion of FIC (Financial Institutional Components) and the growth rate of industry scale; collaborations between large-scale industry models and MNCs (Manufacturing Controllers) boost R&D output; and the choice of funding represents a pricing answer to the question, "Where will the money go after the R&D output increases?" Of course, this judgment is predicated on the premise that the increase in R&D output can actually translate into orders.

II. Why is it important? Where are the scarce assets?

Understanding the current structure of the AI-driven pharmaceutical industry requires a two-layer analytical framework: the capability layer refers to the scientific capabilities of large and basic models, including protein structure prediction, protein and antibody design, virtual cells, genome models, autonomous experiments, and scientific reasoning; the realization layer refers to the actual progress of these capabilities in drug discovery and clinical development.

The misalignment of the three types of subjects: the most capable are furthest from the "medicine".

The first category consists of emerging large-scale model companies (Anthropic, OpenAI, NVIDIA, and Isomorphic Labs under the Alphabet ecosystem, etc.). They are the fastest in terms of capability. Isomorphic Labs' AI drug design engine, IsoDDE, released in February 2026, achieved more than twice the prediction accuracy of AlphaFold 3, released in 2024, on the challenging protein-ligand structure prediction benchmark called Runs N' Poses. In binding affinity prediction, IsoDDE outperformed classic physics-based methods such as Free Energy Perturbation (FEP+, the gold standard physical method in computational chemistry for evaluating molecular binding strength) in three blind tests: FEP+ 4, OpenFE, and CASP16, with only a small fraction of the time and computing power required. The system can also identify new druggable binding pockets, including "hidden pockets" that have remained undetected by the scientific community for over a decade, even when only the amino acid sequence is given and no ligand is specified. This has been validated on the human protein cereblon. Other advancements in the same camp include: OpenAI's GPT-Rosalind, an inference model for life sciences; Evo2, with 40 billion parameters and fully open source, achieving zero-sample interpretation of BRCA1 gene variants with an accuracy exceeding 90%; Xaira's X-Cell, which constructs virtual cells using 25.6 million perturbed single cells; and Ginkgo, in collaboration with OpenAI, running 36,000 real-world experiments, reducing the cost of cell-free protein synthesis reactions by 40%.

However, at the realization level, these companies lack proprietary Phase III drugs. Their efforts to catch up focus on two things: first, mergers and acquisitions and talent poaching. In April 2026, Anthropic acquired Coefficient Bio for approximately $400 million in all-stock, poached John Jumper, a key figure at AlphaFold, and launched the Claude Science research platform in June. Second, deep partnerships with multinational drug manufacturers (MNCs). In January 2026, Nvidia and Eli Lilly announced a maximum investment of approximately $1 billion over the next five years to jointly build an AI drug discovery laboratory. On September 16, Novo Nordisk and Anthropic officially announced their collaboration. The validation on the financing side is equally clear: Isomorphic Labs completed a $2.1 billion Series B financing round in May 2026, led by Thrive Capital, with participation from Alphabet, GV, MGX, Temasek, and others. This is one of the largest private equity rounds in the history of the biotechnology field, having already raised approximately $600 million in the previous year. This type of profit model involves selling token subscriptions, securing access to scientific research, and securing budgets from pharmaceutical companies. The experiments are primarily for feeding back into the model, rather than taking on the risk of new drug failure.

The second category consists of traditional and specialized AI pharmaceutical companies (Schrödinger, Insilico Medicine, Generate, Recursion, Relay, Absci, etc.). Their competitive advantage before large-scale models lies in physical simulation (FEP+), generative chemistry, phenomics, and proprietary data. After large-scale models, the common change is the integration of generative AI, LLM interfaces, and intelligent agents into this competitive advantage, shifting the paradigm from "pure algorithms" to "physical/data + generative AI + intelligent agents." This category has verifiable hard data: Schrödinger's drug discovery collaboration business accounted for 39% of revenue in the first quarter of 2026, a year-on-year increase of 114% to $23.03 million. This segment advances projects in more than 10 therapeutic areas, including oncology, autoimmune diseases, and central nervous system diseases, in collaboration with biopharmaceutical companies, obtaining upfront payments, development milestones, and future sales revenue sharing. Its collaborative asset, the TYK2 inhibitor Zasocitinib (TAK-279), outperformed BMS's marketed drug in a Phase III head-to-head study and is planned to advance its market application in the first half of 2027; Insilico's Rentosertib completed the first dose in China's Phase III trial on September 10, making it the product with the longest clinical progress among AI-native companies; Generate's anti-TSLP antibody GB-0895 is currently recruiting for global Phase III SOLAIRIA-1/2 trials, with a dosing interval of approximately 6 months; Recursion merged with Exscientia for approximately $688 million in all-stock transactions, and REC-4881 entered Phase II. This type of profit comes from subscriptions, open access, and developing their own drugs. The drug development still follows the traditional pharmaceutical model, with profits coming from drug sales.

The third category consists of MNC giants (such as Eli Lilly), who act as demanders. They pursue development along two lines: traditional endogenous development and opening up their proprietary platforms. Eli Lilly's TuneLab is a representative of the latter: the platform launched in September 2025, with its initial model trained on over 500,000 preclinical data points accumulated over 20 years (covering in vivo and in vitro pharmacokinetic and toxicological data). Eli Lilly estimates that the acquisition cost of this proprietary data exceeded $1 billion. The platform is hosted by a third party (Rhino Federated Computing, based on the NVIDIA FLARE framework) and uses federated learning (a privacy-preserving architecture that only shares model updates and does not expose raw data). Participating companies can use Eli Lilly's predictive models to screen candidate molecules and simultaneously feed back experimental results to continuously improve the model. As of September 2026, TuneLab had over 75 participating companies. The motivation of MNCs is completely different from the first two categories—the open platform serves their own drug development, not for modifying or selling models.

The most capable are furthest from the "medicine," while those who need the "medicine" most are furthest from the "model." This misalignment is not a static defect, but rather the driving force behind this round of industry consolidation—it determines two inevitable paths. The first is the upward path, where large model companies and computing power giants supplement their downstream capabilities by acquiring biotech teams and embedding themselves into pharmaceutical company workflows (Anthropic's acquisition of Coefficient Bio and Nvidia's joint lab with Eli Lilly are examples of this). The second is the downward path, where MNCs (Multi-Industry Companies) absorb external data providers into their ecosystems through open platforms (Lilly's TuneLab's integration with Twist and Genscript are examples of this). The connection point between the two paths converges at the same point: the data loop.

From "selling services" to "selling assets": the two paths to realization have diverged.

The misaligned structure determines the direction of industry consolidation, but for specific companies, the path to consolidation is not singular. The interim reports for the first half of 2026 have already made the differentiation between the two models clearly discernible.

Path One: The "Shovel-Selling" Model Represented by XtalPi. In the first half of 2026, the company achieved revenue of 394 million yuan, a year-on-year decrease of 23.9%, with a net loss attributable to the parent company of 252 million yuan and an adjusted net loss of 106 million yuan. More noteworthy is the simultaneous change in both revenue and customer structure. In terms of revenue structure, AI for Science smart solutions revenue reached 194 million yuan, a year-on-year increase of 136.4%, with its revenue share jumping from 15.8% in the same period last year to 49.2%, among which intelligent robot laboratories and intelligent services both maintained rapid growth. In terms of customer structure, the revenue share of the largest customer dropped significantly from 70.6% to 32.9%. The former indicates that the company's revenue engine has shifted from "making medicine for others" to "providing AI4S infrastructure for others"; the latter indicates that this infrastructure is moving from relying on a single customer to replicable, scalable delivery—the simultaneous improvement of these two indicators is more informative than the revenue growth rate itself.

Path Two: The "Asset Selling" Model, exemplified by Insilicon. In the first half of 2026, the company achieved total revenue of US$106 million, a year-on-year increase of 287.2%, of which drug research and development and pipeline development revenue climbed to US$103 million, a year-on-year increase of over 300%, mainly from large down payments from major business development deals that were completed intensively during the period, as well as milestone payments from existing cooperative projects. The total value of the transaction contracts announced by the company in 2026 was approximately US$7.3 billion, and the cumulative total value of signed cooperation agreements since 2021 rose to nearly US$11 billion, with partners including Eli Lilly, Servier, Takeda, SK Biopharmaceutical, and Qilu Pharmaceutical.

XtalPi sold the "right to use R&D capabilities," with revenue in installments, repurchaseable, and without the risk of new drug failure; InSilicon sold the "development rights of pipeline assets," with revenue in blocks, one-time, but with significant potential for asset appreciation.

MNC's question-setting method has changed, but the payment method remains the same.

Among the three types of entities, MNCs are the "problem setters" of order and cooperation structures—they decide whether to purchase large-scale model workbenches or AIDD (AI-assisted drug design) software, whether to use open platforms or single-point pilots, and whether to use decentralized procurement or preferred agreements. From 2025 to 2026, the way MNCs set the problems underwent a clear shift, but the shift was not evenly distributed across the four dimensions.

In terms of collaboration models, the focus has shifted from single-point pilot projects and small contracts to platform-based, multi-target, and open ecosystems, with Eli Lilly's TuneLab being a prime example. Regarding technology procurement, the targets have expanded from AIDD software and CRO services to large model workbench and computational collaboration—Novo Nordisk's purchase of Anthropic's Claude Science, and NVIDIA's joint laboratory with Eli Lilly, are examples of new procurement categories. In terms of external procurement methods, there has been a shift from decentralized purchasing to preferred protocols and data feedback. Twist Biosciences' ordering based on TuneLab's preferred protocol and subsequent data feedback for federated training directly reflects this change. However, the payment structure—low down payment, high milestone biobucks (potential total amount)—remains unchanged, and the characteristic of deferred risk persists.

Breaking down the behavior of MNCs into two layers clarifies things: they treat AI infrastructure as their own asset, essentially "rebuilding" it themselves (the Roche model), while simultaneously treating external capabilities as modules that can be readily integrated and replaced (the Lilly TuneLab model). Both layers of behavior are predicated on data sovereignty—infrastructure can be self-built, proprietary data is not sold externally, and open platforms employ a federated learning architecture rather than directly exchanging data. Understanding this is crucial to truly understanding why "large-scale, labeled wet experimental data" has become the most scarce asset across the entire industry chain.

Who's really putting real money into this game: The truth behind the tiered amounts and down payments in eight transactions.

To answer the question "Which giants have entered the fray?", we need to comprehensively review and compare the major deals publicly disclosed between 2025 and 2026. We have compiled eight representative deals, covering four types: Alphabet ecosystem, large-scale model and computing power giants, Chinese AI platforms, and traditional overseas business development (BD) ventures.

The real change that AI4S brings to the pharmaceutical industry is not that models become smarter, but rather that the bottleneck in the R&D process has shifted. In the traditional model, the bottleneck lies at the very beginning: finding druggable targets and designing active molecules. These two things determine whether a project can enter clinical trials, and therefore, the value is concentrated in these two stages. When generative models reduce the cost and cycle of molecule generation by an order of magnitude, the bottleneck shifts to the validation stage: how to use sufficiently large-scale and standardized experiments to screen out the large number of candidate molecules generated by AI? As a result, the value distribution of the entire industry chain shifts outward, with wet laboratory experiments gaining pricing power due to the scarcity of physical capacity and data accumulation.

III. What to focus on next? The four levels of value outflow.

Four levels: The deeper AI reaches, the sooner it will be repriced.

Dividing the AI pharmaceutical industry chain into four layers based on the direct contribution of AI to the R&D process clearly reveals the differences in pricing power.

The first layer is the model and algorithm layer, including target discovery, molecular generation, and protein structure and affinity prediction. This layer has the strongest AI relevance, but it is also the fastest to be commercialized. The capability gap is being rapidly leveled by open source and competition: Evo2 is open source with 40 billion parameters, allowing any team to perform genome interpretation based on it; Chai Discovery's Chai-2 has become the first zero-sample generation platform to achieve double-digit experimental success rates on completely de novo designed antibodies, and once such capabilities are made public, their premium window will narrow.

The second layer is the wet laboratory and tool layer, including automated experiments (robotic labs), recombinant proteins and key reagents, gene synthesis, model animals, and pharmacological and efficacy evaluation. This is the stage with high AI relevance and high order certainty. There are two pieces of industry evidence: First, Eli Lilly's TuneLab's preferred agreements directly point to data providers like Twist Bioscience and Genscript, indicating that MNC's open platform must rely on external wet laboratory capacity to achieve a closed loop; second, with large model manufacturers entering AI-driven drug development, Genscript's business model has undergone profound changes, and AI-driven protein orders are expected to maintain triple-digit percentage growth in the second half of the year and beyond. The industry logic here is worth elaborating on: the more candidate molecules generated by the model, the more molecules need to be verified experimentally; the more molecules are verified, the larger the sample size for pharmacological, efficacy, and safety evaluation. AI increases the number of shots in R&D, and each shot consumes real experimental resources.

The third layer is the clinical and validation service layer, which requires distinguishing between preclinical CROs and clinical CROs. Preclinical CROs directly undertake the validation needs of AI-generated molecules, sharing the same logic as the second layer, and are more flexible. Clinical CROs primarily focus on patient enrollment, data management, and statistics, with AI's role being more reflected in trial design and data analysis efficiency, resulting in a longer and more indirect transmission chain to R&D volume.

The fourth layer is the manufacturing and end-user segment, including CDMO commercial production, pharmacies, and medical device terminals. AI has the lowest relevance here, serving as a stabilizer in the industry chain rather than a source of flexibility.

Catalyst Outlook for the Next 12 Months

Arranging the catalytic nodes for the next 12 months in terms of their number of years since today reveals a clear two-stage structure.

The near-term focus is on September to December 2026: the EASD European Association for the Study of Diabetes Annual Meeting (September 28 – October 2, Milan), the ESMO European Society for Medical Oncology Congress (October 23–27, Madrid), the SITC Cancer Immunotherapy Society Annual Meeting (November 4–8, Phoenix), the CTAD Alzheimer's Disease Clinical Trials Conference (November 16–19, Boston), and the ASH American Society of Hematology Annual Meeting (December 12–15, New Orleans), with five international academic conferences to be held in succession.

The longer-term focus is on mid-2027: Schrödinger's collaborative asset Zasocitinib (TAK-279) is planned to advance its market application in the first half of 2027; XtalPi guidance indicates that more than 10 pipelines will advance to clinical trials in 2027; and the potential approval window for the first batch of fully AI-designed drugs falls between 2027 and 2028.

Core judgment

Judgment 1: The scarce asset has shifted from model capabilities to data closed loops. Evidence supporting this judgment comes from three levels. At the industry level, the needs of four types of entities converge at the same point: large model manufacturers need real-world scientific questions to validate their model capabilities; MNCs need external capabilities to offset pressure on their self-developed pipelines; AI-native companies need orders and clinical resources to complete the closed loop; and all these stages ultimately require standardized wet laboratory data for implementation. At the funding level, nine of the top ten performing A-share stocks this week are in the wet laboratory and tools sector, indicating that the market has already expressed this judgment through price.

Judgment Two: The entry of industry giants is real, but its value must be measured by initial payments, not headlines. The "real" aspect is reflected in structural changes—major model manufacturers have, for the first time, received R&D budgets from multinational pharmaceutical companies (Novo Nordisk and Anthropic); the MNC open platform has over 75 participants; Eli Lilly's TuneLab's first model is built on over 20 years of experience, more than 500,000 clinical data points, and an investment exceeding $1 billion; Novartis, GSK, BMS, AstraZeneca, Sanofi, Pfizer, and others have all made moves in various forms. These are not declarations of intent, but signed cooperation agreements. This demonstrates both the certainty of the industry's direction and the limited immediate revenue realization.

Risk Warning and DisclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.