Equity Research: Photovoltaic Equipment & Space Infrastructure

Deep Dive: Space Computing Centers – HJT as the Optimal Energy Solution for Orbital Data Centers

Date: January 8, 2026
Analysts: Zhou Ershuang (S0600515110002), Li Wenyi (S0600524080005)
Sector: Photovoltaic Equipment / Aerospace Infrastructure
Key Stocks: Maxwell Technologies (300751.SZ), Gaoce Shares (300751.SZ) [Note: Using English names for international context, referring to Maiwei/Gaoce]

Executive Summary

The convergence of Artificial Intelligence (AI) and aerospace engineering has catalyzed a new paradigm: Space Computing. As global demand for AI training and inference outpaces terrestrial power grid capabilities, the deployment of high-performance computing nodes in Low Earth Orbit (LEO) and Sun-Synchronous Orbit (SSO) is transitioning from conceptual validation to early-stage industrialization. This report analyzes the structural shift towards "Orbital Data Centers," highlighting their disruptive advantages in energy efficiency, cooling costs, and scalability compared to ground-based facilities.

Our core thesis posits that while space computing offers a solution to the terrestrial energy bottleneck, its economic viability hinges on the mass-to-power ratio of the satellite’s energy system. We identify Heterojunction (HJT) silicon solar cells, particularly in ultra-thin (<100μm) and flexible formats, as the optimal technological bridge between current Gallium Arsenide (GaAs) dominance and future Perovskite tandem applications. HJT’s low-temperature processing, radiation resilience, and compatibility with roll-out array structures make it the superior choice for next-generation space photovoltaics (PV).

We recommend investors focus on equipment manufacturers with established overseas client bases and expertise in ultra-thin wafer processing. Specifically, we highlight Maxwell Technologies (Maiwei Shares) as the leader in HJT turnkey equipment, benefiting from its suitability for US-based manufacturing due to lower OPEX and patent safety, and Gaoce Shares for its breakthroughs in 60μm ultra-thin wafer cutting technology, which is critical for reducing launch mass.

Key Takeaways

1. The Emergence of Space Computing: A Response to Terrestrial Energy Constraints

• Paradigm Shift: Space computing involves deploying modular server nodes capable of AI training and inference on LEO/MEO satellites, creating distributed "Orbital Data Centers." This shifts the model from "Sense-on-Space, Compute-on-Ground" to "Sense-and-Compute-on-Space," drastically reducing latency and bandwidth requirements.
• Disruptive Cost Advantage: For a 40MW computing cluster operating over 10 years, the total cost of ownership (TCO) for a space-based system is estimated at $8.2 million, compared to $167 million for a terrestrial counterpart. This ~95% cost reduction is driven primarily by:
  • Energy Costs: Near-zero marginal cost of solar energy in space vs. $140 million in electricity costs on the ground.
  • Cooling: Utilization of the -270°C space environment for radiative cooling eliminates the need for water-intensive cooling towers ($7 million savings) and complex HVAC systems.
  • Deployment: No land acquisition or zoning permits required; scalable via launch cadence.
• Global Race: Major players including China’s Zhejiang Lab & ADASpace, US’s SpaceX, Google (Project Suncatcher), and NVIDIA-backed Starcloud are actively validating this technology. China has already launched 12 AI satellites (Triad Computing Constellation) with stable operation >200 days, while SpaceX and Starcloud have conducted successful test launches of compute-enabled payloads.

2. Energy System Economics: Weight is the Primary Cost Driver

• Cost Structure: The power subsystem (solar arrays + batteries) accounts for ~22% of satellite cost and 20-30% of satellite weight. Given that LEO launch costs remain significant ($1,500–$3,000/kg via Falcon 9), minimizing the mass of the energy system is the single most important factor in improving the unit economics of space computing.
• Array Architecture Evolution: The industry is shifting from traditional Z-fold rigid arrays (used in ISS/Tiangong) to Roll-Out Arrays. Roll-out structures offer higher power-to-weight ratios (>100 W/kg for iROSA vs. ~20 W/kg for legacy ISS wings) and simpler mechanical designs. However, roll-out arrays strictly require flexible, thin-film, or ultra-thin rigid solar cells.
• Technology Selection:
  • Gallium Arsenide (GaAs): Currently the standard for MW-class missions due to high efficiency (~30%) and radiation hardness. However, its prohibitive cost ($1.2 billion/GW) and limited supply chain capacity preclude its use in GW-scale constellations.
  • Silicon-Based PV: The only viable path for mass deployment. While slightly heavier and less efficient than GaAs, silicon costs are 1/6th to 1/3rd of GaAs ($2–3.5 billion/GW).
  • HJT as the Winner: Among silicon technologies, HJT is uniquely suited for space due to:
    1. Low-Temperature Process: Prevents thermal stress and warping in ultra-thin wafers.
    2. Flexibility: Compatible with flexible substrates and roll-out mechanisms.
    3. Radiation Resilience: Demonstrated self-healing properties under electron irradiation.
    4. Tandem Potential: Serves as the ideal bottom cell for future Perovskite/Silicon tandem cells, which promise >30% efficiency.

3. Orbital Dynamics and Market Capacity

• SSO is the Prime Real Estate: Sun-Synchronous Orbit (600–800km altitude) provides near-continuous sunlight (>8,300 hours/year), eliminating the need for large battery buffers and maximizing uptime for high-power data centers.
• Capacity Constraints: Despite the vastness of space, usable orbital slots are finite due to collision avoidance standards.
  • SSO Capacity: At a conservative 30km separation, the SSO band can accommodate approximately 9,616 additional satellite clusters.
  • LEO Capacity: Broader LEO bands (300–2000km) can accommodate ~79,000 additional clusters.
• Scaling Strategies: To maximize revenue per orbital slot, two architectural trends are emerging:
  1. Large-Scale "Motherships": e.g., Starcloud’s concept of a 4km x 4km PV array platform hosting multiple compute modules (5GW capacity).
  2. Swarm Clustering: e.g., Google’s Suncatcher plan to deploy 81–324 satellites in tight formation, acting as a single logical compute unit.
• Demand Implication: A mere 10GW of global HJT production capacity could support ~448 Google-style 324-satellite clusters or 2 Starcloud motherships, indicating that initial PV supply constraints will be a key bottleneck for space computing expansion.

4. Investment Implications: The HJT Supply Chain Opportunity

• US Manufacturing Tailwinds: HJT is uniquely positioned for US domestic manufacturing due to its shorter process flow (4 steps vs. 10+ for TOPCon), lower labor/water/energy intensity, and lack of patent litigation risks (unlike TOPCon and BC technologies). This aligns with the geopolitical trend of reshoring critical tech supply chains.
• Top Picks:
  • Maxwell Technologies (Maiwei Shares): Dominant market share (>70% since 2021) in HJT turnkey equipment. Its latest 1.2GW line reduces CAPEX and OPEX significantly, making it the preferred partner for US/European entrants into space PV.
  • Gaoce Shares: Leader in hard-brittle material cutting. Its proprietary tungsten wire cold-drawing technology enables the mass production of 60μm ultra-thin wafers, a critical enabler for lightweight space PV modules.

Detailed Analysis: The Case for Space Computing

1.1 Why Move Data Centers to Space?

The exponential growth of Large Language Models (LLMs) and AI agents has created an unsustainable trajectory for terrestrial data center energy consumption. By 2030, global data center electricity demand is projected to exceed 3,000 TWh. Grid infrastructure upgrades lag behind this demand, leading to power shortages in key tech hubs (e.g., Virginia, Ireland, Singapore).

Space computing addresses three fundamental limitations of ground-based infrastructure:

A. Energy Cost and Availability
In space, solar irradiance is constant and unattenuated by the atmosphere. The intensity is approximately 1.36 kW/m² (AM0 spectrum), compared to ~1 kW/m² peak on Earth (AM1.5), but with the crucial advantage of 24/7 availability in specific orbits like SSO.
* Ground: A 40MW data center consumes ~350,400 MWh annually. At $0.04/kWh, energy costs are $140 million over 10 years.
* Space: The energy source is free. The only cost is the upfront CAPEX for the solar array (~$2 million for a 40MW equivalent array). This represents a $138 million saving over the asset's life.

B. Thermal Management
Heat dissipation is the second-largest operational challenge for AI clusters. Ground data centers require massive amounts of water for evaporative cooling (approx. 0.5 L/kWh, totaling 1.7 million tons over 10 years for a 40MW plant) and expensive chillers.
* Space: The background temperature of deep space is ~3K (-270°C). Heat can be rejected efficiently via radiative panels without any consumable fluids or moving mechanical parts for cooling loops. This eliminates water usage entirely and reduces cooling-related OPEX to near zero.

C. Latency and Bandwidth Efficiency (The "Edge" Advantage)
Traditional remote sensing satellites downlink raw data to ground stations for processing. This creates a bottleneck:
* Bandwidth Limit: A satellite may generate 500GB of data daily but can only downlink 20GB due to limited window time and spectrum congestion. <10% data utilization.
* Space Computing Solution: On-board AI processors filter, analyze, and compress data in orbit. Only actionable insights or high-value excerpts are downlinked. This increases effective bandwidth utilization by orders of magnitude and enables real-time decision-making for defense, disaster response, and autonomous logistics.

1.2 Global Competitive Landscape

The race for orbital compute is no longer theoretical. Key milestones in 2024–2025 confirm rapid progress:

Source: Company announcements, CNSA, NASA, Dongwu Securities Institute

Strategic Divergence:
* China is leveraging state-led coordination (Zhejiang Lab, CAS) to build integrated constellations that combine remote sensing with edge computing.
* The US is driven by private capital and hyperscaler demand (Google, NVIDIA), focusing on integrating commercial off-the-shelf (COTS) high-performance chips (H100, TPUs) into ruggedized space packages.

Technical Deep Dive: The Photovoltaic Energy System

The economic feasibility of space computing rests entirely on the performance of the photovoltaic (PV) system. Unlike terrestrial PV, where Levelized Cost of Energy (LCOE) is the primary metric, space PV prioritizes Specific Power (W/kg) and Radiation Hardness.

2.1 Technology Comparison: Why Silicon HJT Wins

We evaluated four major PV technologies for space applications: Gallium Arsenide (GaAs), Crystalline Silicon (PERC/TOPCon/HJT), Thin Film (CIGS/CdTe), and Perovskite.

1. Gallium Arsenide (GaAs): The Incumbent with a Ceiling
* Pros: Highest efficiency (~30% for triple-junction), excellent radiation resistance, proven heritage (ISS, Starlink V1/V2).
* Cons: Extremely high cost ($1.2B/GW), heavy substrate (Ge), and brittle nature. Supply chain is constrained by rare material availability (Gallium/Arsenic) and export controls.
* Verdict: Unsuitable for GW-scale deployment. Reserved for niche, high-value military or deep-space missions.

2. Crystalline Silicon: The Mass-Market Solution
Silicon offers a compelling balance of cost and performance. However, not all silicon technologies are equal for space.

Source: Infolink, JPL Radiation Handbook, Dongwu Securities Institute

Why HJT is Superior to TOPCon for Space:
1. Thermal Budget: TOPCon requires high-temperature diffusion steps (>800°C), which induce stress and micro-cracks in ultra-thin wafers. HJT’s entire process occurs below 250°C, preserving the structural integrity of thin (<100μm) wafers.
2. Bifaciality & Light Trapping: HJT’s symmetric structure and transparent conductive oxide (TCO) layers offer better bifacial gain, which is advantageous in the multi-reflection environment of space arrays.
3. Self-Healing Capability: Recent studies by France’s INES show that HJT cells can recover >97% of initial performance after 1MeV electron irradiation when annealed at mild temperatures (80°C). This "self-healing" property extends operational life in LEO radiation belts.
4. Tandem Readiness: HJT is the ideal bottom cell for Perovskite tandems. Its flat, TCO-coated surface allows for direct deposition of Perovskite layers without complex interface engineering required for TOPCon’s textured, insulating oxide surface.

3. Perovskite: The Future, Not Yet Ready
While Perovskite/Silicon tandems promise >30% efficiency and ultra-lightweight characteristics, stability remains a critical hurdle. Space environments involve extreme thermal cycling (-120°C to +120°C) and high-energy particle bombardment. Current encapsulation technologies cannot yet guarantee the 10–15 year lifespan required for commercial constellations. We expect Perovskite tandems to enter commercial space service post-2030, with HJT serving as the foundational technology in the interim.

2.2 Structural Integration: The Rise of Roll-Out Arrays

The mechanical design of solar arrays is undergoing a revolution to accommodate flexible HJT cells.

• Z-Fold (Legacy): Used on ISS and Tiangong. Rigid panels fold like an accordion.
  • Drawback: Heavy hinge mechanisms, low packing density, prone to mechanical failure during deployment. Specific Power: ~20–40 W/kg.
• Roll-Out (Next-Gen): Inspired by tape measures. Flexible solar blankets are wound around a spool and unfurled in orbit.
  • Advantage: Eliminates heavy hinges, maximizes packing density in launch fairings, and achieves specific powers >100 W/kg (as seen in ISS iROSA upgrades).
  • Requirement: Must use flexible cells. Rigid silicon wafers cannot be rolled. This necessitates the use of ultra-thin HJT wafers (60–100μm) bonded to flexible polymer substrates or metal foils.

Case Study: NexWafe & Solestial
* NexWafe (Germany): Produces 70μm ultra-thin HJT wafers using epitaxial lift-off techniques. Secured 250MW contract for LEO satellite power.
* Solestial (USA): Demonstrated continuous production of 60μm HJT cells compatible with roll-out arrays. Their roadmap includes integrating Perovskite top cells to boost efficiency to >30%.

Market Sizing and Orbital Logistics

3.1 Orbital Real Estate: Scarcity in Abundance

While space is vast, useful orbits are congested. Regulatory bodies (ITU, FCC) and industry best practices mandate separation distances to prevent Kessler Syndrome (cascade collisions).

Sun-Synchronous Orbit (SSO): The Premium Zone
* Why SSO? Constant illumination. Satellites in dawn-dusk SSO orbits spend <5% of their time in Earth’s shadow. This minimizes battery size and weight, directly lowering launch costs.
* Capacity Calculation:
* Altitude Band: 600–800 km.
* Inclination: 96–99 degrees.
* Separation Standard: 30 km (aggressive but feasible with active debris removal).
* Remaining Slots: After accounting for ~780 existing satellites, there is room for ~9,616 additional satellite clusters.
* Implication: This finite capacity will drive up the value of orbital slots and encourage high-density clustering (swarms) rather than dispersed single satellites.

Low Earth Orbit (LEO): The Volume Zone
* Capacity: Broader altitude range (300–2000 km) allows for ~79,000 additional clusters at 30 km spacing.
* Trade-off: Higher eclipse frequency requires larger batteries, increasing mass. However, lower launch costs to lower altitudes partially offset this.

3.2 Scaling Models: Motherships vs. Swarms

To maximize the utility of limited orbital slots, two distinct architectural paths are emerging:

1. The "Mothership" Model (Starcloud)
* Concept: A single, massive spacecraft featuring a 4km x 4km (16 km²) solar array.
* Power: ~5 GW.
* Advantage: Centralized thermal management, simplified orbital tracking, economies of scale in power distribution.
* Challenge: Requires next-generation heavy-lift launch vehicles (e.g., SpaceX Starship) for in-orbit assembly or monolithic deployment.

2. The "Swarm" Model (Google Suncatcher)
* Concept: A tightly controlled formation of 81–324 smaller satellites flying in precise coordination.
* Power: ~22–32 MW per cluster.
* Advantage: Redundancy (loss of one node doesn't kill the cluster), easier to launch on existing medium-lift rockets, modular scalability.
* Challenge: Complex inter-satellite linking (laser comms) and formation flying control algorithms.

Production Impact Analysis:
Assuming a standard G12-132 HJT module (2.61 m², ~700W):
* 10 GW of HJT Module Production translates to:
* ~145,000 Starlink V3-sized satellites.
* 448 Google Suncatcher Clusters (324 satellites each).
* 2 Starcloud Motherships.

This illustrates that even modest expansions in terrestrial HJT manufacturing capacity can have a disproportionate impact on space infrastructure deployment. The bottleneck is not just PV production, but launch cadence and orbital slot allocation.

Investment Thesis and Recommendations

The transition to space computing creates a new, high-margin demand vector for the photovoltaic equipment sector. Unlike the terrestrial market, which is currently engaged in a brutal price war, the space PV supply chain values performance, reliability, and weight reduction over lowest-cost-per-watt. This favors companies with advanced technological moats.

4.1 Top Pick: Maxwell Technologies (Maiwei Shares - 300751.SZ)

Investment Logic:
1. Dominant Market Position: Maiwei holds >70% of the global HJT equipment market share. Its technology is the de facto standard for new HJT fabs.
2. US/Europe Expansion Catalyst: As Western nations seek to build domestic PV supply chains (driven by IRA in the US and Net Zero Industry Act in EU), HJT is the preferred technology due to:
* Patent Safety: No active litigation risks (unlike TOPCon/BC).
* OPEX Efficiency: Lower labor, water, and energy consumption aligns with high Western operating costs.
* Compact Footprint: Maiwei’s 1.2GW line reduces factory floor space by 30%, crucial for expensive Western real estate.
3. Space Relevance: Maiwei’s equipment produces the high-quality, low-defect HJT cells required for space applications. Its partnerships with overseas clients position it to supply the non-Chinese space PV supply chain (e.g., Solestial, European aerospace primes).
4. Financial Outlook: With order books robust and ASPs stabilizing, Maiwei is poised to benefit from both the terrestrial HJT replacement cycle and the emerging space niche.

Risk: Geopolitical restrictions on equipment exports to certain jurisdictions.

4.2 Top Pick: Gaoce Shares (300751.SZ)

Investment Logic:
1. Ultra-Thin Wafer Leadership: Gaoce has successfully commercialized 60μm ultra-thin silicon wafer cutting using tungsten wire diamond lines. This is a critical enabling technology for space PV, where every gram of mass saved reduces launch costs by $1,500–$3,000.
2. "Equipment + Consumables" Model: Gaoce sells both the cutting machines and the diamond wire耗材. This recurring revenue model provides stability and high margins.
3. Technological Moat: Tungsten wire cold-drawing technology allows for finer diameters and higher tensile strength than traditional carbon steel, reducing kerf loss and enabling thinner wafers without breakage.
4. Space Application Direct Link: As satellite manufacturers shift from rigid GaAs to flexible silicon, the demand for ultra-thin, flexible-ready wafers will surge. Gaoce is one of the few suppliers globally capable of mass-producing these at scale.

Risk: Competition from other wire manufacturers; potential slowdown in terrestrial PV capex affecting short-term cash flow.

Risks / Headwinds

While the long-term thesis for space computing and HJT adoption is strong, investors must consider the following risks:

1. Policy and Regulatory Volatility:

   • Terrestrial: PV industry remains sensitive to government subsidies, tariff policies (e.g., US Section 301, EU CBAM), and grid connection rules. Changes in trade policy could disrupt supply chains.
   • Space: Orbital slot allocation is governed by international treaties and national regulators. Increased congestion could lead to stricter licensing requirements or liability insurance costs, slowing deployment.

2. Technology Development Risks:

   • HJT Yield: Achieving high yields with ultra-thin (<80μm) wafers is technically challenging. Breakage rates could impact profitability.
   • Space Qualification: The space environment is harsh. Unexpected degradation of HJT cells due to atomic oxygen erosion or proton radiation could shorten satellite lifespans, undermining the economic model.
   • Perovskite Delays: If Perovskite stability issues persist beyond 2030, the efficiency ceiling for silicon-only systems may limit the power density of future satellites.

3. Space Computing Execution Risk:

   • Launch Failures: The model depends on frequent, reliable, and low-cost launches. Any significant setbacks in Starship or other heavy-lift vehicle development could delay constellation deployment.
   • Thermal Management in Vacuum: While radiative cooling is efficient, managing heat spikes from high-density AI chips in a vacuum is unproven at scale. Overheating could lead to compute throttling or hardware failure.
   • Market Adoption: Enterprise customers may be hesitant to migrate critical workloads to space due to perceived security risks or latency variability until the technology is fully matured.

4. Competition from Terrestrial Alternatives:

   • Advances in nuclear fusion, next-gen fission (SMRs), or grid-scale storage could alleviate terrestrial power constraints, reducing the urgency for space-based solutions. However, given the 10–20 year timeline for these technologies, space computing retains a near-to-mid-term advantage.

Rating / Sector Outlook

Sector Rating: OVERWEIGHT

We maintain an Overweight rating on the Photovoltaic Equipment sector, specifically targeting companies with exposure to HJT technology and ultra-thin wafer processing. The emergence of space computing adds a high-growth, high-margin layer to the traditional terrestrial PV demand story.

Investment Horizon: 12–24 Months.

Key Catalysts to Watch:
1. Launch Cadence: Successful deployment of Starcloud’s full compute cluster and Google’s Suncatcher prototypes (2027–2028).
2. HJT Cost Parity: Confirmation of HJT achieving cost parity with TOPCon in terrestrial markets, driving wider adoption and economies of scale that benefit space supply chains.
3. Regulatory Clarity: International agreements on orbital traffic management and space debris mitigation.
4. Technical Milestones: Public demonstration of >30% efficient Perovskite/HJT tandem cells in orbit.

Investment View: Strategic Allocation

For institutional investors, we recommend a barbell strategy within the PV equipment space:

1. Core Holding: Maxwell Technologies (Maiwei Shares).

   • Role: Capture the broad secular trend of HJT replacing PERC/TOPCon globally. Benefit from its dominant market share and expanding presence in Europe/US.
   • Alpha Driver: Space computing contracts provide optionality and margin expansion potential beyond the commoditized terrestrial market.

2. Satellite Holding: Gaoce Shares.

   • Role: Pure play on the "lightweighting" trend. As both terrestrial modules (to reduce BOS costs) and space satellites (to reduce launch costs) move to thinner wafers, Gaoce’s technology becomes indispensable.
   • Alpha Driver: Monopoly-like position in tungsten wire cutting equipment for ultra-thin applications.

Conclusion:
Space computing is not merely a sci-fi concept; it is an engineered response to the physical limits of terrestrial energy and bandwidth. The pivot to HJT solar technology is the linchpin of this transition. By investing in the equipment makers that enable this shift—specifically those who can produce high-efficiency, lightweight, and flexible PV solutions—investors can capture value from both the ongoing energy transition on Earth and the nascent industrialization of space.

Appendix: Detailed Financial and Technical Data

A. Cost Comparison: Ground vs. Space Data Center (40MW, 10-Year TCO)

Source: Starcloud, IEA, Dongwu Securities Institute Estimates

B. HJT vs. TOPCon: Operational Metrics for US Manufacturing

C. Orbital Capacity Calculations (SSO 600-800km)

• Earth Radius (R): 6,371 km
• Orbit Radius (r): R + Altitude
• Circumference (C): $2 \pi r$

Scenario 1: 50km Separation (Conservative)
* Average Circumference: ~44,420 km
* Layers (600-800km, 50km spacing): 5 layers
* Total Track Length: ~222,100 km
* Max Satellites: $222,100 / 50 = 4,442$
* Existing Satellites: ~780
* Remaining Capacity: 3,662

Scenario 2: 30km Separation (Aggressive/Active Debris Removal)
* Layers (600-800km, 30km spacing): 7 layers
* Total Track Length: ~311,900 km
* Max Satellites: $311,900 / 30 = 10,397$
* Existing Satellites: ~780
* Remaining Capacity: 9,617

Note: These figures represent "slots" for large clusters or motherships, not individual smallsats. A single Google Suncatcher cluster occupies one such slot.

Final Thoughts

The intersection of AI and Aerospace represents one of the most significant capital allocation opportunities of the next decade. While the headlines focus on chip designers and LLM developers, the physical infrastructure—power and cooling—will determine the pace of growth. Space offers a unique solution to the energy bottleneck, and HJT PV is the key that unlocks this potential.

Investors should look beyond the traditional terrestrial PV cycle and recognize the emerging Space-PV Industrial Complex. Companies like Maxwell Technologies and Gaoce Shares are not just selling equipment; they are providing the foundational tools for the orbital economy. Their technological leadership in HJT and ultra-thin wafer processing positions them as critical enablers of this new frontier.

We advise institutions to accumulate positions in these leaders during any market dips related to short-term terrestrial PV oversupply concerns, as the long-term value proposition from space computing remains underpriced by the broader market.

Disclaimer:
This report is prepared by Dongwu Securities Co., Ltd. for institutional clients only. It does not constitute an offer to sell or a solicitation of an offer to buy any securities. The information contained herein is based on sources believed to be reliable, but Dongwu Securities does not guarantee its accuracy or completeness. Past performance is not indicative of future results. Investors should conduct their own independent research and consult with financial advisors before making investment decisions.

Analyst Certification:
The analysts named in this report certify that they have accurately described the views and opinions expressed in this report. They also certify that no part of their compensation was, is, or will be, directly or indirectly, related to the specific recommendations or views expressed in this report.

Contact:
Dongwu Securities Research Institute
No. 5 Xingyang Street, Suzhou Industrial Park, Suzhou, China
Zip Code: 215021
Website: http://www.dwzq.com.cn