The shift from competence trust to relationship trust ultimately forms a resilient and trustworthy AI ecosystem that drives sustainable development. Here's why.

The Trump administration proposed the “Stargate” plan, aiming to strengthen the US’s global competitive edge by constructing a massive artificial intelligence (AI) infrastructure. Powerful computing capabilities and vast data accumulation have driven the commercialisation of powerful AI models like those from Google, Meta, Anthropic, and OpenAI.

Meanwhile, the emergence of DeepSeek has undoubtedly brought fresh momentum to the industry, not only achieving technological breakthroughs but also showcasing new possibilities in its approach. DeepSeek has chosen a "low-cost, high-performance, open-source collaboration" artificial intelligence path, opening up foundation model architectures and lowering development barriers to enable more enterprises, developers, and research institutions to participate in the co-construction of the global AI ecosystem.

Reflecting on the “surprise” of DeepSeek, several key insights emerge:

1. The competitive landscape of AI in China, the US, and Europe: Shaping the future through technological cooperation and competition

The surprise of DeepSeek reflects a new phase in the global AI competition. Historically, the US has maintained a dominant position in high-performance computing and has implemented export restrictions on advanced chips to curb China’s AI development.

However, rather than falling behind, China has forged an efficient and innovative AI development path despite these challenges. This is not only a testament to China’s technological resilience but also serves as a valuable reference for the global AI community.

Today, China, the US, and Europe arguably form a tripolar structure in AI leadership, while the rise of open-source models can be an opportunity to enable broader participation in AI innovation worldwide – across all other regions. Openness not only can accelerate the adoption of AI technologies, if done right, but can also foster a new paradigm where global tech competition is increasingly interwoven with opportunities for cooperation.

2. The influence of capital markets: From technology-driven to technological ecosystem-driven

Capital markets have always been a catalyst for innovation. Despite short-term fluctuations, AI breakthroughs are now arguably driving a fundamental shift in global capital allocation as new value creation opportunities, as well as disruptions of businesses and industries, emerge.

Unlike traditional industries that rely heavily on significant resource consumption, AI innovation has the potential to introduce a more efficient growth model. Moreover, unlike the first phase of capital-intensive foundation AI model development investments (from OpenAI to Anthropic, Mistral, and others), capital markets can now make investments in smaller, highly specialised, and innovation-driven companies that contribute to the advancement of AI-driven innovation and the broader AI ecosystem. This shift will not only unlock new opportunities for financial markets but will also lay a solid foundation for the sustainable development of AI technologies.

In the future, the value of “AI native” companies may no longer be measured solely by computing power but rather by algorithmic efficiency, ecosystem scalability, and the depth and breadth of real-world applications they enable.

3. Inclusive innovation: Making AI model development and use accessible to all

DeepSeek’s open-source and cost-efficient approach can be a game-changer for individual developers and small enterprises, and will likely trigger the creation of many small, innovative, cost-efficient powerful AI model developments across the world.

In the early stages of AI development, the high technological barriers meant that only large or very well-funded tech companies could afford the costs of training and optimizing advanced AI models. This is now changing.

Moreover, the new wave of low-cost, high-efficiency open-source tools can lower the entry barriers for using powerful AI models, enabling a broader range of participants to contribute to AI innovation and applications. This new phase of democratisation of AI can not only further accelerate business and societal progress but also reinforce the principle that technology should be inclusive rather than exclusive to a privileged few. Innovation should not be a luxury—it should be a driving force that elevates society as a whole.

So given where we are, how can, or perhaps should the AI journey continue? We believe that AI ecosystem building is not only a critical direction for future developments but also a necessary pathway toward achieving powerful AI that we can then leverage to solve major challenges and progress ahead.

Of course, this requires building global trustworthy AI ecosystems. To accomplish this, strong governance practices need to be set up, as also argued before. Governance, especially when stakes are high, should start with a powerful mission, clear vision, a fair inclusive process, and a focus on building trust.

Mission: AI as an engine for progress

Demis Hassabis’ Nobel Prize in Chemistry speech this year highlighted the amazing promises AI has for the future of humanity. As he noted in the past, and in line with ideas from his MIT advisor Tomaso Poggio, solving the problem of intelligence will enable us to then use that (stronger) intelligence to help solve all other problems. 

However, AI innovation should not be an inefficient energy-intensive race but rather an exploration of new application scenarios that enhance productivity and well-being.

Beyond data processing and intelligent decision-making, AI is set to revolutionise industries such as healthcare, education, and industrial manufacturing. Furthermore, AI is poised to become a key enabler in expanding humanity’s frontiers -- from low-orbit satellite communication and industrial internet to low-altitude economy and virtual reality. By transcending the limitations of the physical world, AI will unlock new creative possibilities, allowing human ingenuity to flourish across vast new dimensions.

Vision: Large-scale global AI adoption

Value creation requires not only AI model performance but also AI adoption. Unless AI is used at scale and globally, its full value potential will not be reached. Lowering the barriers to AI adoption requires, among others, reducing the cost and complexity of AI applications, and making them adaptable to different industries, contexts, and use cases.

Two key next steps should focus on lightweight AI applications and on scenario-based customization of powerful AI models. By adopting a modular and open-source approach, organisations can build AI systems tailored to their specific needs without incurring high training and customisation costs. Whether in intelligent healthcare applications, education, industrial inspection, or financial risk control, AI solutions should be seamlessly embedded into various products, services, and business scenarios in a cost-effective and efficient manner.

Lowering the barriers to AI adoption will drive wider technology diffusion and maximize the potential of this powerful general-purpose technology for the world.

Fair process and inclusion through ecosystems

AI should serve not only large enterprises in advanced economies but also empower small and medium-sized businesses, and individual entrepreneurs across all nations to integrate AI at low cost, accelerating its use across diverse applications. Moreover, it is not only about providing powerful models and about competition, but also about fostering ecosystems that enable value co-creation.

A robust AI ecosystem requires a multi-tiered partnership network that integrates infrastructure providers, industry-specific AI solution providers, and the broader AI community.

Taking DeepSeek as an example, at the infrastructure level, strengthening collaboration with domestic computing power providers, such as Huawei Ascend, will be crucial in advancing local and global AI computing capabilities. At the application level, working closely with AI solution providers across industries like healthcare, manufacturing, and finance will accelerate the integration of AI into real-world business scenarios, ensuring that AI technologies generate tangible value.

At the ecosystem level, leveraging the power of the global open-source community can enhance technological innovation and transparency and drive innovation diffusion. By encouraging developers to contribute algorithms and optimise models, the ecosystem can become more inclusive, enabling a broader range of stakeholders to participate in and benefit from AI advancements. Through layers of collaboration, an AI ecosystem can be cultivated that is not only technologically advanced but also sustainable and widely accessible.

Trust at the core

Trust is the foundation of technology adoption. It is also at the core of partnerships, collaboration and competition which are necessary to drive innovation.

There is plenty of research on how to build and maintain trust, particularly when it comes to building ecosystems. For example, one of us worked on a framework of gradual trust building through different stages: from goodwill trust to competence trust and ultimately to relationship trust.

Establishing goodwill trust requires enhancing transparency through open-source initiatives and cost-effective strategies, ensuring that enterprises and developers can access AI technologies without unnecessary barriers. This fosters confidence in an open and mutually beneficial AI ecosystem.

As the ecosystem matures, competence trust must be built by continuously optimising model performance, expanding multimodal capabilities, and ensuring AI security and interpretability. These efforts help industries recognise the reliability and value-creating potential of AI technologies.

Beyond competence trust, long-term stability in partnerships is essential to forming relationship trust. By fostering enduring collaborations with computing power providers, industry partners, and the open-source community, AI ecosystem governance can be strengthened, promoting global cooperation. This shift from competence trust to relationship trust ultimately forms a resilient and trustworthy AI ecosystem that drives sustainable development.

We may be witnessing a fundamental shift in the AI development paradigm, and with that also in how our AI world may look in the near future: from capital-intensive and relatively exclusive closed AI model development, to more efficient and open AI that can enable widespread adoption which will lead to significant value creation and changes in the world.

The next challenges are no longer only technological. The future of AI will not be determined solely by computing power or algorithms—it will be shaped by our ability to establish globally trusted AI ecosystems and to ensure that AI becomes a shared innovation driver for all of society.

The story of the AI era has merely just begun.

Theodoros Evgeniou is a professor of Technology and Business at INSEAD. Ke Rong is a professor at the School of Social Sciences at Tsinghua University, China, where he also serves as the director at the Institute of Economics as well as deputy dean at the Institute for Global Industry. He is a member of one of China's top think tanks, providing policy advice to the central government, and is part of the Expert Network of the World Economic Forum.