Brokerages say the market can recover, focus on investment opportunities in physical AI, solid-state batteries, innovative drugs, and more

Main body

I. Major news

On October 23, NASA announced the latest arrangements for the "Artemis 3" mission. Under the plan, the mission will send four astronauts into low Earth orbit in 2027, where they will successively conduct rendezvous and docking tests and other tests with lunar landing system test vehicles developed by Blue Origin and others.

II. Latest broker assessments

Huafu Securities: The market can recover; avoid short-term disturbances

The benign cycle in emerging industries continues to consolidate. Sectors related to high-tech industries and new quality productive forces are steadily strengthening their role in driving economic growth, growth drivers are continuously iterating and optimizing, and the endogenous momentum of economic transformation is steadily accumulating. Overall, the economy accelerated quarter on quarter in the third quarter and business conditions picked up at the margin, which to some extent broke the earlier pessimistic market expectations of weak fundamentals and also laid the fundamental groundwork for subsequent corporate earnings delivery and market valuation repair.

After this round of adjustment, the valuations of A-share core assets have fully digested pessimistic expectations, and the allocation value of the equity market has gradually become prominent. At present, many broad-based and sector indices have fallen to historically low percentiles, some growth sectors are already at extremely low levels for the past year, and value sectors such as financials and cyclicals are likewise in historically bottom areas. The negative factors that previously weighed on valuations have been reflected in share prices to a large extent, downside room is limited, and the risk-reward ratio has clearly improved. Current valuation levels have already fully priced in external disturbances and short-term sentiment concerns. Going forward, as long as earnings stabilize at the margin and liquidity expectations recover moderately, equity assets will have upward repair elasticity. From a structural perspective, high-quality names with strong earnings certainty and reasonable valuations are expected to usher in a relatively good allocation window.

Policy and capital continue to build momentum; external disturbances will not change the medium- to long-term direction, and both the capital and policy fronts are simultaneously accumulating positive signals. On the one hand, broad-based ETFs have seen net inflows for multiple consecutive trading days, and institutional funds are clearly positioning against the trend; on the other hand, since 2026, the upper limits of buyback and shareholding increase plans announced by companies on the Shanghai and Shenzhen stock exchanges have totaled more than 100 billion yuan, with industrial capital expressing recognition of their own value with real money. On the policy side, the Ministry of Finance issued the first batch of 150 billion yuan in special government bonds to inject capital into central financial institutions, also opening up space for credit extension and long-term asset allocation.

As third-quarter report disclosures enter their final stage, the inflection point of earnings repair is expected to be further confirmed. Coupled with the continued entry of incremental funds, the market may gradually digest external disturbances and return to the main line of fundamental pricing. It is advisable to avoid short-term sentiment interference, avoid panic selling, focus on high-quality names with a good match between earnings and valuation, and maintain rational and patient positioning.

III. Broker sector opportunities

Physical AI, represented by world models, can be regarded as an intermediate station on the path from LLMs to superintelligent robots. LLMs excel at language understanding, knowledge retrieval, and task decomposition, and have already achieved commercialization loops of considerable scale in white-collar work domains such as coding, general office work, finance, and law; of course, they are still at an early stage of penetration.

The next stop for AI is to enter the physical world. Physical AI also needs to understand three-dimensional space, object states, contact relationships, and dynamic constraints, and to predict the possible outcomes of different actions. World models learn environmental states and how they change over time and with actions, building an internal representation of the environment that robots can use for simulation, enabling language instructions to be further transformed into perception, prediction, planning, and control. Under this framework, VLA maps visual and language inputs to robot actions, while world models provide future-state prediction and outcome evaluation for action selection.

Embodied intelligence scenarios, data, and models are evolving in sync, and high-value, low-repetition tasks are expected to be the first to form commercial loops. Standardized, highly repetitive tasks are already covered by general-purpose or specialized equipment; non-standard tasks such as disaster rescue, maintenance of complex equipment, flexible manufacturing, and home care better reflect the cross-scenario value of humanoid robots; near-term deployment will still mainly be in To B scenarios where paying entities and return calculations are clearer. The training system is gradually forming a three-track integration of real machines, simulation, and human video: real-machine data calibrates physical errors, simulation data expands environments and failure-recovery trajectories, and human video provides task semantics and skill priors.

At present, manufacturing constraints are concentrated in dexterous hand and forearm assembly, tactile sensor reliability, tooling precision, calibration efficiency, and supplier consistency. As more robots enter workstations and warehouses, data from real tasks, failure trajectories, and human takeovers will flow back into model training, simulation, and evaluation systems, driving reliability improvements and scenario replication. Industrial value will also extend from core components to automated assembly, inspection and calibration, data governance, simulation training, and deployment and operations platforms. Specific related names include Tesla, AgiBot, SWANCOR Advanced Materials, and Unitree Robotics; the North American chain includes Kesen Technology, Silin Zhiqu, Fusai Technology, and Daimei Co.; other robot chains include Lingyi iTech, Pan Asia Micropermeable, Orbbec, and Leaderdrive.

Southwest Securities: Solid-state battery pilot production capacity is landing in concentrated fashion, and industry development has entered a critical window

Against the backdrop of surging global new energy demand and liquid lithium battery performance approaching physical limits, solid-state batteries, with advantages in energy density, safety, and lifespan, have become the inevitable direction for next-generation power batteries. However, their industrialization is still constrained by multiple factors including materials, interfaces, processes, and costs. They are expected to enter the pilot testing and small-batch mass production window in 2026-2027, and investment opportunities will first materialize in equipment and materials. Demand-side pressures are forcing a technological leap, and liquid lithium batteries have reached their "ceiling." Global sales of new energy vehicles and demand for power batteries have entered a TWh-level expansion phase, significantly raising comprehensive requirements for range, safety, and fast charging.

Solid-state batteries are a "system-level solution," but they will not happen overnight. The industrialization rhythm is clear, and investment opportunities follow a "chronological order." From 2025 to 2026, semi-solid/quasi-solid-state batteries will be the first to land, with the core beneficiaries in materials validation + equipment orders; from 2026 to 2027, pilot lines will be built intensively, marking a year of engineering validation, with a production line construction cycle of about 8 to 10 months and equipment orders released in advance; from 2028 to 2030, technology routes will gradually converge, entering large-scale commercialization. Overall, solid-state batteries represent the long-term evolution direction of the power battery system, and their strategic value has been widely recognized by industry and policy levels, but in the short term they remain in a stage of "validation-driven rather than volume-driven." The key to grasping solid-state battery investment opportunities lies not in betting on a single technological breakthrough, but in understanding the industrialization rhythm, following the phased dividends, and making forward-looking arrangements along the path of "materials—equipment—batteries—applications."

Specifically for companies in the industrial chain, on the battery side, they include CATL, Sunwoda, EVE Energy, Ganfeng Lithium, etc.; on the materials side, they include Shanghai Xiba, Orient Zirconic, Sanxiang Advanced Materials, Sinocera Material, etc.; on the equipment side, they include Lead Intelligent, Kexin Electromechanical, Hangke Technology, Honggong Technology, etc.

Guosheng Securities: China's innovative drugs step onto the global core stage, with multiple pipelines reaching key data readouts

Global participation of China's innovative drugs has increased, and LBA selections hit a new high. The 2026 European Society for Medical Oncology (ESMO) Annual Congress was held in Madrid. At this congress, the international influence of China's innovative drugs rose significantly, with a total of 28 Chinese studies selected for Late-Breaking Abstracts (LBA), 4 of which even entered the plenary session, accounting for one third. This shows that Chinese pharmaceutical companies are moving from "following" to "racing," with key pipelines fully entering the data readout period and continuously contributing "Chinese data" to global oncology.

IO evolution is upgrading from monoclonal and bispecific antibodies to diversified "cytokine fusion proteins/trispecific antibodies." The IO2.0 era has expanded from the single track of PD-1/VEGF bispecific antibodies to diversified route competition of "bispecific antibodies + cytokine fusion proteins + trispecific antibodies."

At the same time, the RAS target, once regarded as "undruggable," has entered an accelerated development period. Companies such as GenFleet Therapeutics and Hengrui Pharmaceuticals appeared on the same stage, demonstrating Chinese pharmaceutical companies' rapid catch-up and differentiated layouts in difficult targets such as Pan-RAS and KRAS G12D. Specifically for related targets, they include Hansoh Pharmaceutical, Kelun-Biotech, Innovent Biologics, CSPC Pharmaceutical Group, etc.

(Source: National Business Daily) Mid-article embedded advertisement