The VC funds-oriented government in Hefei city serves as a development model for many local governments in China. Since the central government’s issuance of the “Guidance on the Standardized Establishment and Operation of Venture Capital Guidance Funds” in 2008, it has become a very common phenomenon for local governments to conduct venture capital investments through policy-guided funds.

After the release of the action plan known as “Made in China 2025”, the number of local policy-guided fund programs surged, reaching a peak in 2016 with 556 new funds established and a total investment amount of 1.7886 trillion yuan. One key reason for the rapid development of China’s new energy automotive industry is the strong support of venture capital, in which local governments’ policy-guided funds have played a crucial role.

In China, for the most competitive industries and economically vibrant cities, local governments often adopt the Six-Force Model for Industrial Policy and Competitive Advantage, aimed at promoting the development of high-tech industries. This involves enhancing factor inputs, building infrastructure, reducing transaction costs, creating industrial clusters, expanding market scale and encouraging industry competition. The Hefei model is unique in that the application of this six-force model runs parallel to a  better leverage and enhanced input for venture capital as one of the advanced production factors, equally developing related risk control measures to ensure the sustainability of government venture capital investments.

Backed by venture capital investment, Hefei has achieved the localization and industrialization of several disruptive technologies, leading to a revolutionary transformation of the local economic structure. Hefei has developed a VC funds-oriented model of government combining elements from both the American and Japanese styles of, respectively,  entrepreneurial and developmental government. We owe this emerging approach to the fact that, within just 20 years, Chinese economy has transitioned from a developing high-value-added manufacturing model, characteristic of Japan’s development model, to focusing on disruptive technologies.

 

In 2004, the Chinese government clearly proposed a goal for  “independent innovation.” From that point onwards, the Chinese government began to adopt industrial policies similar to those implemented by Japan in terms of developmental government. However, in just 10 years, Chinese economy has fully entered various high-tech industries and started to compete on par with the United States. At this stage, the government has naturally begun to engage in venture capital investments.

The VC funds-oriented government in Hefei has pioneered a high-tech industry development model, with state-owned capital taking the lead and social capital providing necessary support. A series of equity investments have facilitated the introduction of new energies-based vehicle OEMs and component enterprises into the local area, thus creating a brand-new industrial ecosystem and leveraging the cluster effect to drive greater market expansion in the region.

At the same time, Hefei government has developed a systematic decision-making process and various measures in terms of risk management and control. Nowadays, government officials widely acknowledge the risk of inaction that comes with excessive caution or fear of taking risks, while sustaining that overly reckless risks can still harm positive outcomes. Since the investment funds of the VC funds-oriented government are state-owned assets, ensuring the sustainability of the government’s investments in high-tech industries does requires ensuring the safety of these state-owned asset investments.

The risk management and control measures adopted by Hefei’s VC funds-oriented government align with the common practices in the venture capital industry. While they cannot ensure success in every government investment, they have still led to a significantly overall reduction in investment failures.

Nowadays, AI video generator Sora is still far from its promoters’ claim that it represents a major breakthrough in artificial general intelligence (AGI), nor has it further widened the technological gap between China and the US in the AI field. Rather, it is merely a product of brute force computation backed by massive capital investment, without much technical innovation. Likewise, its societal benefits do not outweigh the costs.

Large-scale language model training is a game characterized by heavy capital investment where players burn through both money and energy; ,  not at all energy-efficient and environmentally friendly. This is where we need to conduct a cost-benefit analysis of the commercial and social applications of large models, for us to determine whether the benefits outweigh the costs, and how both are distributed.

When conducting such an analysis, we should consider those externalities that have not been factored into the costs, such as the environmental damage and systemic risks actually borne by the public, as well as any potential benefits that it may bring. Currently, the applications that are attracting widespread media and public attention are merely aimed towards individual users (consumer-facing), such as text, audio and video generation. However, these types of applications are likely to be short-lived.

Artificial intelligence was originally a major public project in the 20th century, but it has been ruthlessly privatized ever since. Huge huge economic benefits have been restricted to a select circle at the top of the pyramid. Sadly, those who have been exploited continue to be amazed by the magic that tech giants show them, indulging in the “AI fantasy” and envisioning AI systems as detached intellects that are independent of their creators, infrastructure, and the world at large in absorbing and generating knowledge.

Policy guidance and institutional intervention are key methods to ensure that AI development aligns with human purposes, with humanism serving as the fundamental principle guiding our discussions on the goals of AI advancement. However, some people believe that we should abandon the anthropocentric approach, suggesting that we shouldn’t limit the boundless potential of AI with human constraints. This includes the acknowledgment that AI may eventually surpass human intelligence, attain self-awareness, and possibly even exert control over or eliminate humans.

This view is akin to the arguments of certain extreme environmentalists who advocate “eliminating humans to protect the Earth,” which is a value choice that goes beyond the realm of rational discussion —as well as one that most of humankind will not accept. The humanist approach thus requires that we properly position AI as a tool, only developing those AI technologies that may assist human activities.

Humans do not need to prove their “usefulness” to justify their need for existence. Immanuel Kant stated in his Formula of Humanity that humans are ends in themselves, never mere means —this should be the starting point of humanist thinking. On one hand, even if AI can already perform most tasks that require human intelligence, humans should use institutional design to ensure that the quality of human life does not worsen. On the other hand, and more importantly, we need to reflect on a direction worth pursuing for AI —promptly and resolutely.  

Next, from a humanist perspective, I believe the generative AI development path represented by OpenAI has gone in the wrong direction. Tools like ChatGPT and Sora, which are generative AI, are doing things that the average human creator is able to do, if they are willing. However, due to a lack of human-directed intentionality and environmental sensitivity, the content they generate can only meet the needs of entertainment for superficial purposes and cannot be used for producing content that pursues authenticity, rigor, and sincerity.

Currently, the digital development between the US and other regions in the world presents a “hub-and-spoke” structure. In this scheme, we can observe that the US occupies the center of the information order (hub) and is highly involved in the digital development of the satellite regions (spokes). At the same time, the US has massively plundered the public resources of these satellite regions, impacting and reshaping their political and economic structures as a direct result.

Firstly, it must be stated that the US is in a leading position in the global data center field, with a much larger number and scale of data centers compared to other countries. Data show that in 2023, the number of data centers in the US mainland exceeded 5,000, while the total number in Germany, the UK and China remained below 2,000. In addition, US Internet companies own more than half of the world’s hyper-scale data centers. This indicates that the US possesses massive computing power resources. As a result, the cost of directing data flows to the US for processing remains relatively low, even in regions outside the US. For example, data from many European companies is first processed in US-based data centers before being used by local users.

Most micro and small enterprises are opposed to Europe’s strengthening of cross-border data regulation, as this would significantly increase their operating costs. This cross-border information transmission model inevitably reminds people of the telegraph communication era, when French military communications relied on British telegraph lines, with information first transmitted to London before reaching Paris. Even in the telephone communication era, communication between Côte d’Ivoire and its neighboring country Ghana had to be routed through London.

Next, when it comes to the consideration of factors such as data sovereignty, we can observe that US Internet companies have long deployed data centers globally. This has further consolidated their core position in the information field, endowing them with overwhelming computing power advantages. As a result, the US has core competitiveness in technologies such as cloud computing and artificial intelligence. Although regions like the EU have begun to strictly regulate cross-border data flows, their domestic digital development still relies in no negligible ways on US Internet companies. This has formed the hub-and-spoke structure in the global information flow order, with the US occupying the central position.

In recent years, the deployment of data centers by US companies has been synchronized with the digitalization of local societies in satellite regions, including the cloud migration of the public sector and private enterprises to cover areas such as telecommunications, energy, finance and urban governance. Taking Europe as an example, the European Commission has expressed hopes that by 2030, 75% of European companies will adopt cloud computing services. However, this goal is highly dependent on US Internet companies. Data shows that US Internet companies like Amazon, Microsoft, and Google dominate 70% of the cloud computing market in Europe. Telecom operators in France and Germany strive to enhance their data sovereignty, but the build-up of their sovereign cloud platforms still relies on Google and Microsoft’s respective cloud technologies.

The dependence of developing countries on this hub-and-spoke structure is even more evident. In Nigeria, 80% of public health data is stored on the cloud servers of companies like Google, Amazon and Microsoft. Local scholars have pointed out that this setup is only benefitting the profitability and data acquisition power of Internet companies, without ever improving Nigeria’s ability to store public data. In 2022, the Indian conglomerate Adani Group signed a cloud computing contract with Google to migrate its core applications and data to Google Cloud. The African bank NedBank also announced the transference of its user database to Microsoft for full management.

It is evident that the digital development of both developed and developing countries relies on technologies such as cloud computing. By leveraging their global data center deployments, US Internet companies have effectively made various regions’ digital development dependent on them. Their significant computing power has suppressed the development of local data center companies. In 2018, the Indonesian government attempted to relax restrictions on foreign companies building data centers to promote economic development, but encountered opposition from local business organizations, concerned that the move could prove harmful to national information sovereignty and the interests of local businesses.

In summary, US Internet companies, leveraging their global data centers as a springboard, have heavily intervened in the digital development of the satellite regions. These, in turn, have opened up their territory to these companies while also making their own contribution to the critical resources of our digital age —computing power and data.

More importantly, by deploying data centers in satellite regions, American Internet companies are able to expand their product and business operations into those local markets. Taking Europe as an example, the Amsterdam data center market has seen an average annual growth rate of over 18% from 2014 to 2021, forming a first-tier market in Europe together with Frankfurt, London, and Paris. The massive infrastructure resources have propelled the market expansion of US Internet companies across Europe and even Africa.

This reveals the other side of the hub-and-spoke structure of global digital development, where US Internet companies have heavily exploited the public resources of the satellite regions, thereby generating profound political and economic impacts.