The global computer industry is entering a transformative phase in 2025, driven by rapid advancements in artificial intelligence (AI), enterprise software innovation, financial technology (FinTech), autonomous driving, and domestic digital infrastructure. This mid-year outlook explores key technological shifts, market dynamics, and growth opportunities shaping the sector's evolution across multiple verticals.
Market Performance and Macroeconomic Context
From the beginning of 2025 through June 26, the computer industry (as classified by CITIC) posted an 8.8% gain, outperforming the CSI 300’s 0.3% rise and the ChiNext Index’s 1.3% decline. Despite this relative strength, valuations remain selective—software sub-sectors have seen modest re-rating, while hardware and equipment segments trade at comparatively lower multiples.
As of mid-2025, the sector’s trailing P/E ratio (TTM, overall method excluding negative values) stands at 47.45x. Investor sentiment is increasingly influenced by real-world commercialization progress rather than speculative momentum, with a growing focus on revenue visibility and sustainable business models.
👉 Discover how leading tech platforms are leveraging AI for real-time market insights and execution.
Core Growth Drivers Across Key Sub-Sectors
AI Models and Tools: Generative AI Fuels Enterprise Adoption
Generative AI continues to accelerate its penetration into enterprise workflows, supported by steady improvements in reasoning capabilities and multimodal integration. Global leaders like OpenAI and Google have pushed the boundaries with advanced models such as GPT-o3-mini and Gemini 2.0 Flash, while domestic players including DeepSeek, Doubao (ByteDance), and Kimi (Moonshot AI) are rapidly closing the performance gap.
Token usage has surged as a key indicator of adoption depth. Google’s monthly token processing jumped from 9.7 trillion in April 2024 to 480 trillion in April 2025—a nearly 50x increase. Similarly, Doubao reported a daily token volume of 16.4 trillion by May 2025, up from 120 billion in May 2024. Notably, demand is shifting beyond content creation toward high-value use cases: code generation, scientific research, and enterprise automation now dominate usage patterns.
This trend is directly translating into cloud revenue growth. In Q1 2025:
- Microsoft Azure Cloud revenue grew 33%, with AI contributing 16 percentage points.
- Google Cloud revenue rose 28.1% year-over-year.
- AWS increased by 16.9%.
Domestically, DeepSeek’s technical innovations—particularly in cost-efficient inference—have catalyzed broader model adoption. Its open-source breakthroughs have enabled enterprises with strict data governance requirements (e.g., state-owned enterprises and financial institutions) to deploy large models on-premises securely.
AI Agents: From Assistant to Autonomous Executor
AI agents represent the next evolution in human-machine interaction. Solutions like Manus and Operator demonstrate end-to-end task automation—from parsing unstructured resume files to generating structured Excel outputs with ranked candidates—while learning user preferences over time.
These agents operate asynchronously in the cloud, freeing users from device dependency and enabling complex data analysis workflows without local compute constraints. Use cases in education planning, investment analysis, and travel coordination validate their scalability.
For B2B applications, AI agents offer measurable ROI through process automation, decision support, and workflow optimization—transforming AI from a novelty into a core productivity engine.
Enterprise Software: Structural Shifts Amid Weak Demand Recovery
While macroeconomic conditions remain subdued, enterprise software demand shows resilience due to long-term digitization imperatives. The manufacturing PMI briefly dipped below 50 in April 2025 but has largely stayed above the contraction threshold since October 2024, signaling fragile recovery.
Despite soft demand, major ERP vendors like Kingdee International maintain stable revenue trajectories due to:
- Persistent under-digitization among Chinese enterprises.
- Rising executive recognition of IT’s role in cost reduction and competitiveness.
- Geopolitical pressures accelerating domestic substitution efforts.
Domestic Substitution Gains Momentum
In R&D-intensive industrial software—such as EDA (Electronic Design Automation), CAD, and CAE—the U.S.-China tech decoupling has intensified self-reliance initiatives. U.S. export controls on advanced EDA tools (e.g., GAAFET architecture) have made compliance non-negotiable for Chinese semiconductor firms.
As a result:
- Domestic EDA player Huada Empyrean acquired Xindar Technology and Synopsys Semiconductor to expand its digital design portfolio.
- Wondertide Software enhanced its CAx suite with AI-powered features like intelligent dimensioning and automated drafting.
- Local ERP adoption is gaining traction among SOEs and central government agencies aiming to replace SAP and Oracle systems incrementally.
According to IDC, China’s core industrial software market will grow from RMB 318.6 billion in 2024 to RMB 765 billion by 2029 (CAGR: ~19%). The AI-enhanced segment is projected to grow even faster at a CAGR of 41.4%, with generative design and predictive quality analytics leading investment priorities.
AI Integration Reshapes ERP Architecture
ERP systems are evolving from process-centric platforms to intelligent decision engines powered by AI. Short-term applications include:
- Predictive analytics: Forecasting cash flow, inventory needs, or customer churn using historical and real-time data.
- Process automation: Automating invoice processing, procurement approvals, or HR onboarding tasks.
- Personalized services: Tailoring marketing offers or after-sales support based on behavioral insights.
Longer term, AI agents could enable "command-driven" ERP interactions. For example:
A procurement officer instructs: “Order 5,000 units of component X from approved suppliers within budget.” An AI agent verifies stock levels, selects vendors via pre-negotiated contracts, initiates purchase orders, schedules delivery, and triggers payment upon receipt—all autonomously.
Kingdee’s苍穹 Agent Platform 2.0 supports such multi-agent collaboration with modules for financial reporting (Jinyue Report), business intelligence (ChatBI), recruitment, travel management, and knowledge search—each tailored to high-frequency enterprise scenarios.
Financial IT: Stablecoins Lead Regulatory Innovation
Stablecoins are emerging as a critical bridge between traditional finance and digital asset ecosystems. With a global market cap exceeding $250 billion—dominated by USDT and USDC—the regulatory landscape is maturing rapidly.
Regulatory Milestones in Hong Kong and the U.S.
- Hong Kong: The Stablecoin Ordinance, effective May 30, 2025, establishes a comprehensive licensing regime for fiat-backed stablecoin issuers. The government also plans to regularize tokenized government bonds and promote broader asset tokenization across commodities and renewables.
- United States: The Senate passed the GENIUS Act on June 17, mandating full reserve backing (in USD or short-term Treasuries), tiered oversight, and enhanced transparency—aimed at preserving dollar dominance while mitigating systemic risks.
These frameworks signal institutional confidence and pave the way for wider adoption in cross-border payments (where stablecoin transaction volume surpassed Visa in 2024), supply chain finance, and digital economy platforms.
Traditional financial institutions are responding swiftly:
- Guotai Junan International became the first Chinese-owned broker in Hong Kong licensed for full virtual asset trading services.
- Third-party payment leaders like Newland Payment, GLENOVA (New Guodu), and LAKALA are expanding globally with localized acquiring solutions and cross-border settlement products.
AI Transforms Financial Operations
DeepSeek’s open-source models are being integrated into banking, insurance, and securities workflows:
- ICBC deployed DeepSeek privately within its “ICBC Smart Surge” model matrix, enabling over 200 intelligent applications across customer service, risk control, and investment research.
- Xinhua Insurance launched personal AI assistants in its mobile app using DeepSeek-R1/V3 models.
- GF Securities introduced DeepSeek-powered client services on its institutional platform “GF Smart Hub.”
Meanwhile, core banking system modernization under the national IT innovation (Xinchuang) initiative is accelerating. Top-tier brokerages like Everbright Securities have begun migrating to next-generation trading systems with decoupled compute-storage architectures—creating new opportunities for vendors like Dingdian Software.
Operating Systems: HarmonyOS Expands Beyond Mobile
HarmonyOS is no longer just a mobile alternative—it’s becoming a full-stack ecosystem spanning PCs, wearables, smart homes, automotive systems, and industrial IoT devices.
Key milestones in early 2025:
- Huawei unveiled its first native HarmonyOS laptop—the MateBook Pro—powered by the self-developed Kirin X90 chip and HarmonyOS 5.
- By June 2025, over 40 device types ran HarmonyOS natively.
- Over 9,000 apps participated in system-level co-development; more than 30,000 HarmonyOS-native apps were available.
- Huawei launched the “Harmony Starlight Program,” allocating RMB 100 million to support developer innovation in distributed AI and native intelligence.
Third-party vendors are actively building on the ecosystem:
- Swelllink (subsidiary of iSoftStone) released SwanLinkOS PC Edition based on OpenHarmony, supporting x86 platforms.
- Ronghe Software, iSoftStone, and Pactera are deepening their roles as strategic partners in enterprise application development.
With over 1 billion HarmonyOS-equipped devices expected by year-end, the platform is poised to challenge legacy ecosystems—not just domestically but globally.
👉 Explore how blockchain-based identity systems are enhancing secure access across digital platforms.
Computing Infrastructure: Domestic AI Chips Rise
Global hyperscalers continue heavy investment in AI infrastructure:
- Microsoft: $16.7B capex (+53% YoY)
- Google: $17.2B (+43%)
- Amazon: $25B (+68%)
- Meta: $12.9B (+102%)
NVIDIA dominates with Blackwell-based GB200 NVL72 clusters accounting for 70% of data center compute sales in Q1. However, export restrictions on high-end AI chips (e.g., H20 series) have created space for domestic alternatives.
Huawei’s Ascend 910C now exceeds NVIDIA A100 performance in BF16 compute (780 TFLOPS vs. 624) and offers larger memory capacity (128GB). Its CloudMatrix 384 system delivers 300 PFLOPS—1.7x faster than GB200—deployed in major smart computing centers in Gui’an, Hohhot, and Wuhu.
Other developments:
- Cambricon reported improved supply chain stability with inventory up RMB 980 million QoQ.
- H3C launched the S12500AI switch using DDC architecture, supporting over 70K GPU clusters with >100% bandwidth improvement.
- Baidu introduced its Kunlunxin Super Node powered by domestically designed AI chips.
With government-backed smart computing centers coming online nationwide in Q3 2025, domestic chipmakers stand to benefit from large-scale procurement with high localization ratios.
Autonomous Driving: Mass Adoption Accelerates
ADAS penetration is rising sharply:
- L2+ systems reached 5.2% penetration.
- L2++ (with NOA/parking assist): up to 12.6%, +5.3pp YoY.
- Entry-level L1/L0 configurations dropped below 6%.
Notably, affordability is expanding access:
- ADAS controller shipments in the RMB 100k–200k price band surged over 2,000% YoY in early 2025.
- BYD’s mass rollout of intelligent driving features across mid-tier models is forcing industry-wide upgrades.
Investment focus remains on:
- Autonomous driving chips – Horizon Robotics and Black Sesame Technology lead domestic innovation.
- Tier-1 suppliers – Desay SV Automotive benefits from scale economies.
- Export-ready integrators – Ruimai Technology gained early-mover advantage in Europe under GSR Regulation EU 2019/2144 for commercial vehicles.
Black Sesame is also advancing into humanoid robotics using its A1000/C1200 chip series—the same architecture powering smart vehicles—for perception-decision-control loops in next-gen robots.
Energy IT: Grid Investment & Market Reform Drive Growth
National grid capex hit RMB 702 billion in 2024 (+20%), with Southern Grid spending RMB 127 billion (+12%). This robust investment underpins stable demand for energy IT vendors.
The release of Document No. 136 in February 2025 marked a turning point: it mandates full market participation for renewable energy generation. This opens new revenue streams for:
- Virtual power plants (VPPs)
- Electricity trading platforms
- Demand response systems
A national VPP policy issued in April sets targets:
- 20 million kW adjustable capacity by 2027
- 50 million kW by 2030
These policies remove previous barriers to grid integration and create viable business models for flexibility providers.
Frequently Asked Questions (FAQ)
Q: When will AI applications deliver measurable ROI for enterprises?
A: Many organizations are already seeing returns through automated reporting, customer service chatbots, and code generation tools. Full workflow transformation may take 1–3 years depending on legacy system complexity.
Q: Is domestic AI chip technology competitive globally?
A: While still behind NVIDIA’s top-tier offerings, Huawei’s Ascend series now matches or exceeds A100-class performance. With strong government backing and local ecosystem alignment, domestic chips are gaining share rapidly in China.
Q: How will stablecoin regulation impact fintech innovation?
A: Clear frameworks from Hong Kong and the U.S. reduce uncertainty and attract institutional capital. Compliance-first stablecoins will likely become standard rails for cross-border settlements and programmable finance.
Q: Can HarmonyOS succeed outside China?
A: Initial success will depend on developer adoption and enterprise partnerships. Its distributed architecture gives it an edge in IoT scenarios where seamless device interoperability matters most.
Q: What drives growth in energy IT despite economic headwinds?
A: Grid modernization is mission-critical infrastructure with stable funding. The shift toward energy market liberalization creates new monetizable services around forecasting, trading, and grid balancing.
Q: Are cybersecurity risks increasing with AI adoption?
A: Yes—risks include data poisoning, misinformation generation, model theft, and adversarial attacks. Enterprises must adopt zero-trust frameworks and invest in AI-specific security layers like prompt injection detection.
👉 Learn how secure digital wallets are evolving alongside AI-driven threat detection systems.