Business Value is Evolving Again, Thanks to AI

China’s “hundred models war” has effectively ended with DeepSeek’s rise. As the global competition for AI supremacy shifts from developing large models to focusing on practical applications, the once-popular “flash-in-the-pan” products by Baidu—such as Wenxin Yiyan and AI Six Little Tigers—have seen their influence wane. Developers are now increasingly prioritizing expanding the added value derived from their applications. With AI models increasingly seen as open resources and digital “public goods,” true innovation lies in how they are applied.

A value-driven approach to AI application development—one that emphasizes inclusiveness, social good, and ethical principles—is crucial in guiding the responsible evolution of AI within the global business ecosystem. The future of AI will depend not only on technological breakthroughs but also on how well it balances business value with social impact.

Can “Model Equality” Unlock the Next Wave of Demand?

The concept of “model equality”—making AI models more accessible and cost-effective—has the potential to significantly expand demand. When DeepSeek shook U.S. and European tech markets, Microsoft CEO Satya Nadella cited the “Jevons paradox,” an economic principle stating that efficiency gains often lead to higher total consumption as costs fall and new use cases emerge . By lowering costs and increasing applicability, new technology tends to stimulate demand. Nadella predicted that AI’s rising convenience and efficiency will drive its rapid spread, transforming it into a ubiquitous “commodity.”

AI, in this context, acts as an “efficiency engine.” Rising usage will fuel demand for data centers, Nvidia chips, AI talent, and even new energy sources like nuclear power. Similarly, more efficient chatbots could uncover new applications and expand usage beyond current expectations. The performance of DeepSeek’s models—comparable to larger U.S. models despite operating with fewer resources—suggests that data centers may become more, rather than less, productive.

This outlook is bullish on AI, emphasizing not only efficiency but also rapidly growing demand. Where high costs hinder adoption, efficiency gains—achieved through cheaper, open models—lower barriers. Where innovation raises expectations—as in healthcare or pharmaceuticals—generative AI could absorb a larger share of social spending.

While early applications—marketing copy, image generation, knowledge bases, code assistants—boost efficiency, they struggle to show quantifiable gains to revenue growth, market share or customer satisfaction. Consequently, companies are scrutinizing the actual return on investment (ROI) from their AI initiatives.

The Application Layer: AI’s Real Economic Engine

The focus is shifting toward real-world applications that generate tangible benefits—such as automating government tasks, enhancing public safety, and improving enterprise productivity. Integrating AI in social and economic systems is increasingly viewed as essential for delivering meaningful long-term impact.

Although DeepSeek has yet to achieve breakthroughs in model capability, its low cost and open-source nature are expected to foster economies of scale at the application level. This could, in turn, boost overall productivity and extend economic output boundaries, positioning AI as a key driver of economic growth.

DeepSeek’s widespread adoption is actively promoted by China’s government. For instance, Shenzhen’s Futian district employs 70 AI civil servants to automate tasks traditionally reliant on manual labor—such as generating meeting notes and documents. Shenzhen officials have used DeepSeek to analyze surveillance footage, successfully locating 300 missing persons. Similarly, in Meizhou and Guangzhou, local authorities are deploying DeepSeek for citizen hotlines, showcasing its practical societal benefits.

From Technological Enablement to Business Reshaping

AI is transitioning from a general-purpose tool to offering specialized solutions—such as autonomous AI agents capable of performing complex tasks independently. This evolution is fundamentally reshaping organizational workflows and decision-making processes. Innovations like Retrieval-Augmented Generation (RAG) are empowering smarter enterprise applications and knowledge management systems.

AI agents are rapidly becoming “killer applications.” Platforms such as Adept and MultiOn are exploring ways for AI to directly operate software systems to automate complex tasks. Future AI agents—endowed with genuine understanding, planning, and execution abilities—are poised to become “digital employees” working across marketing, customer service, R&D, coding, supply chain management, and strategic decision-making. This promises both efficiency gains and structural business transformation.

Existing large models are increasingly directed toward industry-specific applications, aiding experts in fields like medical image analysis, drug discovery, and financial risk management. However, businesses must remain cautious of AI agents’ limitations and avoid over-reliance. Strengthening human-machine collaboration and cultivating resilient relationships between AI agents and human staff is vital. AI possesses immense potential for both internal collaboration and external communication. Internally, it can accelerate document processing, improve knowledge management, streamline communication, and facilitate multilingual teamwork.

AI for Social Good: The Real Breakthrough

AI has already achieved remarkable results in advancing research, healthcare, education, environmental protection, energy, and public services. Like the internet’s evolution from information tool to platform economy, generative AI’s true transformation will rely on its ability to deliver widespread social value and foster social trust.

Prioritizing societal needs through AI applications in healthcare, education, environmental protection, and rural development can foster greater social trust and inclusiveness. These efforts can lead to broader acceptance and responsible deployment of AI, ensuring that its benefits are accessible to diverse communities.

Fostering “AI inclusiveness” can reveal the unexplored “blue oceans” in markets that have so far been overlooked by conventional business logic. AI can be put to work addressing issues an aging society will struggle with, or come up with solutions for rural revitalization schemes. This will broaden our understanding of AI’s value for social good.

Generative AI’s Critical Challenges

The advent of efficient open-source models like DeepSeek has reshaped the generative AI sector, offering significant opportunities for innovation around the globe—particularly in China. However, technological progress alone does not guarantee commercial success. Overcoming obstacles such as technological readiness, costs, user acceptance, model scalability and sustainability are all obstacles that must be overcome.

A “social value first” strategy can guide us in developing AI agents that carry out qualitative change, and reshape the global business ecosystem responsibly while leading the new generation in its industrial revolution. For this to work, industry must embrace a long-term vision beyond short-term profits and commit to technology for good.

This article was originally published in Chinese on Financial Times Chinese.