HAITONG INT'L Zhang Yidong: AI is expected to drive innovative drugs from performance improvement toward a revaluation of R&D asset value

HAITONG INT'L Zhang Yidong issued a research report stating that innovative drugs are expected to become one of the directions with substantial growth potential and valuation elasticity in the AI application stage, and their uniqueness lies in the fact that the value brought by AI may far exceed current-period revenue and profit improvement. For most industries, AI mainly realizes value through efficiency improvement and profit growth in existing businesses; while the value of innovative drug companies depends on the potential commercialization value of pipelines under development, and their valuation depends not only on current-period earnings, but also on factors such as future expected peak sales of drugs, the scale of cash flows contributed, the timing of realization, and the probability of realization. AI R&D efficiency improvement is expected to improve the input-output ratio of existing projects and may improve the conditions for realizing some cash flows; large-scale new drug discovery is expected to expand potential R&D assets and commercialization space, driving upward revisions to long-term growth expectations. The latter does not necessarily shorten cash flow duration, but may increase the expected value of companies' future cash flows. The main views of HAITONG INT'L are as follows: 1. Why has the high-level upward move in US Treasury yields not significantly suppressed the performance of innovative drugs? In the past few years, the medium-term trend of the innovative drug index and US Treasury yields has shown a clear negative correlation. Innovative drug companies have the characteristics of long R&D cycles, later realization of cash flows, and strong reliance on external financing. Especially for early-stage Biotech companies that have not yet commercialized, their value mainly comes from expectations for the long-term cash flows of drugs under development, and thus they are relatively sensitive to changes in US Treasury yields and discount rates. Since 2026, US Treasury yields have continued to rise, but both Chinese and US innovative drug indices have achieved gains, and the valuation pressure brought by rising rates has not become the dominant factor affecting sector performance. As of October 14, 2026, the Hong Kong Stock Connect Innovative Drug Index rose about 10.1%, and the Nasdaq Biotechnology Index rose about 20.4%; during the same period, the US 10-year Treasury yield rose from 4.15% to 5.19%, accumulating an increase of about 104bp. From the 250-period rolling correlation coefficient, the correlation coefficient between Hong Kong innovative drugs and US Treasury yields rose to +0.11, and the original negative correlation weakened significantly; the US NBI rose to +0.83. 2. Cash flow duration repricing: the internal logic of innovative drugs strengthening under high rates 2.1 Valuation mechanism: how does cash flow duration repricing affect the sensitivity of innovative drugs to US Treasury yields? The sensitivity of innovative drug valuations to US Treasury yields mainly depends on the distribution of future cash flows and the level of discount rates. As pipelines under development enter the commercialization stage, licensing transactions and R&D efficiency improvements occur, some cash flows are expected to be realized earlier, or their probability of realization improves, thereby changing the expected distribution of future cash flows and alleviating the valuation pressure brought by rising rates. When expectations for future cash flows of innovative drugs undergo substantive improvement, the positive impact of industry fundamentals may exceed the valuation pressure of rising rates and become the main contradiction affecting innovative drug valuations. With the continuous advancement of technological progress, international cooperation such as out-licensing and co-development of products, companies are expected to increase near-term cash flows, or push forward the timing of realization of future cash flows, enabling company value that originally relied heavily on long-term expectations to obtain clearer cash flow support, thereby bringing valuation elasticity that exceeds the negative impact of interest rate fluctuations. A simplified DCF model can intuitively demonstrate this valuation mechanism: the valuation increase brought by improved cash flow distribution may exceed the discounting pressure caused by rising rates. 2.2 Innovative drugs are not a broad rally: commercialization, BD and AI all point to improved cash flow expectations The US pharmaceutical market shows clear divergence, and new drug commercialization, M&A transactions, and AI drug discovery have become important DRIVE factors for the rise of the innovative drug sector. As of September 30, 2026, the Biotechnology ETF (XBI) and the Pharmaceutical ETF (IHE) had risen 29.42% and 19.51% year-to-date, respectively, significantly higher than the 10.07% gain of the broad Healthcare ETF (XLV), while the Medical Devices ETF (IHI) fell 17.35%. The strong performance of the innovative drug sector is mainly related to key clinical data readouts of innovative technology therapies demonstrating commercialization potential, sales of already commercialized products exceeding expectations, and multinational pharmaceutical companies (MNCs) continuing to acquire high-quality pipelines. The penetration of AI technology into the pharmaceutical industry is becoming a long-term variable for improving the return on investment of the pharmaceutical industry, catalyzing efficiency improvement and cost reduction in traditional pharmaceutical methods in the short term, and in the long term is expected to explore new biological mechanism discoveries and address more unmet disease needs. The rise of the Hong Kong Stock Connect Innovative Drug Index is likewise not a broad rally, but is mainly driven by a small number of companies with relatively clear cash flow realization paths. Among the 41 constituent stocks, only 16 companies rose and 25 fell; under a simulated scenario with unchanged constituents and holdings, the top ten positive contributors together contributed about 19.98 percentage points, while the remaining 31 companies together dragged down about 9.48 percentage points. Index returns are mainly concentrated in three types of companies: first, innovative drug companies whose profitability is improved by product commercialization and global BD; second, R&D and production service companies with order growth and strong delivery capabilities; third, tools and experimental service platforms expected to benefit from growth in AI R&D demand. Overall, the market places greater emphasis on the actual value realization capability and future cash flow growth expectations of innovative drug companies. 3. Global supply-demand resonance in innovative drugs drives continued repricing: patent cliffs release demand, new technologies expand supply 3.1 Patent cliff pressure continues to emerge, and demand from large global pharmaceutical companies for external acquisition of innovative assets is released. IQVIA estimates that by 2030, more than US$230 billion of existing biopharmaceutical sales will face the risk of losing market exclusivity. The demand for multinational pharmaceutical companies to supplement innovative assets through M&A, licensing and joint development continues to grow. In the first half of 2026, global biomedical M&A amounted to about US$130 billion, already close to the full-year 2025 scale of about US$133 billion. 3.2 Supply side: new technologies and high-value innovative drugs continue to emerge, expanding the global innovative drug market space. The market space for innovative drugs depends not only on existing disease demand, but is also driven by new technology breakthroughs and treatment regimen improvements. Innovative technology therapies such as antibodies, ADCs, peptides, and small nucleic acid drugs continue to expand the treatment boundaries of disease areas such as oncology, metabolism, autoimmune diseases, and cardiovascular diseases, driving the continuous emergence of new drugs with clinical competitiveness and commercialization potential. According to IQVIA data, the share of novel therapeutic technologies in globally initiated oncology clinical trial projects increased from about 11% in 2016 to about 33% in 2025; the average annual number of global oncology new active substances (NAS) launched increased from about 16 in 2015-2019 to about 26 in 2020-2024. The development of new technologies is driving the upgrading of the innovative drug R&D structure and providing richer choices for large global pharmaceutical companies to acquire high-quality innovative assets. 3.3 Chinese innovative drugs realize value along the dual paths of product going global and R&D assets going global, with more companies crossing the break-even point. On the one hand, companies such as BeOne Medicines Ltd. and Legend Biotech share in global market space through overseas product launches and sales growth; on the other hand, Chinese Biotech companies realize the international value of R&D assets through License-out. The sustained revenue growth brought by products going global, together with the increase in licensing revenue brought by R&D assets going global, jointly drive more Chinese innovative drug companies to cross the break-even point, with profit improvement spreading from a few leaders to more companies. Taking the top 20 companies in the Hang Seng Hong Kong Stock Connect Innovative Drug Index as a sample, the number of companies with positive net profit attributable to parent was 11 and 10 in 2023 and 2024, respectively, increasing to 14 in 2025 and further reaching 17 in the first half of 2026, with the proportion of profitable companies rising from 50% in 2024 to 85% in the first half of 2026. 4. AI reshapes the innovative drug R&D paradigm: opening up a new round of valuation repricing space 4.1 AI technology is expected to reverse the anti-Moore's Law status of the pharmaceutical industry, change the R&D economics of innovative drugs, and expand growth boundaries Innovative drug R&D has the characteristics of "long cycle, high investment, and high failure rate," and R&D efficiency has long constrained the industry's return on investment. According to Deloitte data, the average R&D cost for a single new drug from discovery to launch at large global pharmaceutical companies rose from US$2.23 billion in 2024 to US$2.67 billion in 2025. AI technology, through computational prediction, molecular design and experimental screening, is expected to improve the efficiency of R&D resource use, reduce capital and time investment in some R&D stages, and improve the industry's long-term mismatch between R&D input and output. The more profound value of AI lies in expanding the boundaries of new drug discovery and driving drug R&D from project-by-project exploration to a large-scale, platform-based model. With the continuous advancement of AI models in the pharmaceutical and scientific research fields, the improvement of AI's ability to propose scientific hypotheses, and the widespread application of 24-hour automated laboratories, disease mechanisms, potential targets and molecular modalities that were difficult to explore in the past are expected to become new R&D opportunities, making large-scale new drug discovery possible. This means that AI can not only improve the R&D economics of existing projects, but is also expected to expand the developable pipelines and future commercialization opportunities of innovative drug companies, further opening up the industry's long-term growth space. 4.2 AI spawns R&D "shovel sellers" and new business models, opening up growth space in the innovative drug industry chain Similar to how computing infrastructure was the first to benefit from the development of the AI industry, experimental validation and data platforms in innovative drug R&D are expected to become the first segments of the AI drug discovery industry chain to achieve commercial returns. As AI drug R&D accelerates, demand for high-quality data and large-scale experimental validation is expected to continue growing, and the industry chain will thereby generate new orders, revenue and business models. First, high-quality data and testing platforms are expected to benefit from growing demand for AI model training and scientific research exploration. Gene sequencing, molecular testing and clinical data platforms can expand their original testing and data production capabilities into model training, data licensing and R&D services. Illumina and Tempus AI provide relevant business cases. Second, in vitro experimental validation platforms are expected to be the first to benefit from the new experimental validation demand brought by improved efficiency of AI-designed molecules. Businesses such as gene synthesis and protein expression benefit from the growth in demand for AI candidate molecule design and experimental validation. GenScript's AI-enabled drug discovery-related business revenue doubled year-on-year in the first half of 2026; Twist Bioscience's new orders contributed by AI drug discovery-related customers in fiscal year 2025 accounted for about 38% of the full-year order increment, indicating that AI R&D demand has begun to actual business opportunities. Third, in vivo experimental platforms are expected to benefit from the extension of drug validation demand to subsequent R&D stages. After candidate molecules pass preliminary screening, animal models are still needed to validate in vivo efficacy, safety and mechanism of action. Platforms with animal models, antibody R&D and proprietary data capabilities are expected to expand R&D cooperation and technology licensing businesses. Fourth, the combination of AI and automated experiments is expected to spawn new platform-based businesses such as "molecule factories," model training and deployment. Relevant companies can not only provide single experimental services, but may also charge continuously through integrated R&D platforms, dedicated model training and localized deployment, driving revenue sources to expand from traditional R&D projects to technical services and R&D asset cooperation. 5. Innovative drugs are expected to become a high-elasticity track for AI applications, releasing valuation elasticity beyond performance growth 5.1 Historically, a new round of repricing of innovative drugs has not necessarily been premised on a decline in US Treasury yields, and industry breakthroughs have repeatedly driven pharmaceutical assets through interest rate pressure The US Biotech market in 2013 showed that an improvement in the innovative R&D boom may drive a repricing of expectations for the entire sector. From the end of April to the end of 2013, the US 10-year Treasury yield rose from about 1.70% to 3.04%, an increase of about 134bp; during the same period, the NBI rose about 65.6% for the full year. That year, new drug approvals, IPOs and industry transactions were active, driving a reassessment of the commercial value of innovative therapies. The GLP-1 market in 2023 showed that in a high-rate environment, the market's reward for commercialized cash flows may be more concentrated. The US 10-year Treasury yield once reached 4.98%, yet Eli Lilly's share price still rose about 59% for the full year, while the total returns of XBI and XLV during the same period were about 7.6% and 2.1%, respectively. Mounjaro sales increased from US$483 million in 2022 to US$5.163 billion in 2023, reflecting the important support of product commercialization ramp-up for corporate valuation. 5.2 Beyond cost reduction and efficiency improvement: AI is expected to drive innovative drugs from performance improvement toward a revaluation of R&D asset value Innovative drugs are expected to become one of the directions with substantial growth potential and valuation elasticity in the AI application stage, and their uniqueness lies in the fact that the value brought by AI may far exceed current-period revenue and profit improvement. For most industries, AI mainly realizes value through efficiency improvement and profit growth in existing businesses; while the value of innovative drug companies depends on the potential commercialization value of pipelines under development, and their valuation depends not only on current-period earnings, but also on factors such as future expected peak sales of drugs, the scale of cash flows contributed, the timing of realization, and the probability of realization. AI R&D efficiency improvement is expected to improve the input-output ratio of existing projects and may improve the conditions for realizing some cash flows; large-scale new drug discovery is expected to expand potential R&D assets and commercialization space, driving upward revisions to long-term growth expectations. The latter does not necessarily shorten cash flow duration, but may increase the expected value of companies' future cash flows. Therefore, a new round of valuation repricing for innovative drugs is not necessarily premised on a decline in US Treasury yields. High rates will still exert discounting pressure on long-term cash flows. Products going global and global licensing cooperation have improved the visibility of near-term revenue and cash flows, and AI is expected to further improve the economics and potential growth space of subsequent R&D pipelines. When the positive revision of industry fundamentals is sufficient to offset the negative impact of rising discount rates, innovative drugs may still achieve valuation repricing in a high-rate environment. Looking ahead, if US Treasury yields decline, innovative drugs are expected to usher in a dual resonance of AI-driven improvement in R&D economics and a decline in discount rates, driving simultaneous increases in long-term growth expectations and valuation levels. Risk warnings: 1. AI technology's improvement in new drug R&D and clinical success rates is less than expected; 2. Clinical progress is less than expected; 3. New drug R&D failure, and review and commercialization progress are less than expected; 4. Fluctuations in upfront payments and milestone realization in global License-out cooperation, with relatively high one-time gains; 5. New orders, collections and capital expenditure returns of AI R&D platforms are less than expected; 6. Intensified industry competition landscape; 7. Risk of changes in AI and pharmaceutical industry-related policies or review and approval systems.