In today's rapidly evolving technology landscape, Artificial Intelligence (AI) has emerged as a force to be reckoned with, sparking profound changes globally. With its unique intelligent features and vast potential applications, AI presents unprecedented opportunities and challenges for human society's development, driving transformations across various industries and profoundly impacting society. Exploring the boundless value of AI not only deepens our understanding of this technology but also offers new avenues for future innovation and development.

However, the rapid advancement of AI technology also brings a series of ethical and responsibility issues. For instance, how can we ensure the fairness and transparency of AI decisions to avoid algorithmic bias and discrimination? How can we protect personal privacy and data security to prevent information leakage and abuse? These questions require in-depth consideration and resolution. Therefore, establishing AI ethical norms, strengthening supervision, and promoting self-discipline are crucial for ensuring the healthy and sustainable development of AI.

Nevertheless, the restrictions imposed on AI ethical issues have hindered AI's progress. This is particularly evident in the increasingly fierce international technological competition. In response, the United States, as a leader in AI, has recently initiated a series of actions.

01

Origin of the US AI 'Manhattan Project'

Recently, the U.S.-China Economic and Security Review Commission (USCC) recommended to Congress that it 'establish and fund a project similar to the Manhattan Project, dedicated to achieving General Artificial Intelligence (AGI) capabilities sooner.' The Manhattan Project was a massive collaboration between the U.S. government and the private sector during World War II, leading to the development and production of the first atomic bomb.

According to reports, the USCC's 793-page report also suggested that Congress allow the executive branch to contract and fund AI, cloud computing, and data center companies, integrating the efforts of the government, businesses, and research institutions to accelerate AGI research and ultimately establish the 'U.S. as a leader in the field of AGI.'

Currently, common AI tools like ChatGPT and AlphaFold are limited to specific tasks such as processing language material and calculating protein structures, categorized as 'Artificial Narrow Intelligence' (ANI). In contrast, 'Artificial General Intelligence' (AGI) is seen as having the potential to revolutionize human society and explode human productivity because it can autonomously learn and think like humans, perform diverse tasks across multiple domains without human intervention. For example, AGI could be used to develop complex investment strategies and research methods for treating incurable diseases, continuously iterating and upgrading itself.

In 2015, before OpenAI existed, its predecessor was Y Combinator's AI lab. Sam Altman, then the president of YC, invested $10 million of his own money and began recruiting talent to spin off OpenAI. When pitching his idea to Elon Musk, Altman referred to the upcoming OpenAI as the 'Manhattan Project for AI.' In an email to Musk, he wrote, 'Have you ever thought about YC (a startup incubator) starting a 'Manhattan Project' for AI? My feeling is that we could start with 50 top researchers and, through some kind of nonprofit model, make the technology belong to the world. But if it's successful, those involved will also be well-compensated.'

Shortly after Trump's election, an official AI infrastructure blueprint from OpenAI was leaked. Titled 'Blueprint for an American AI Infrastructure,' the document covered AI economic zones, nuclear power projects, and privately funded government projects. The Blueprint aimed to establish AI economic zones co-created by state and federal governments to 'incentivize states to expedite permitting and approval of AI infrastructure.' The company planned to build new solar panels and wind farms and reactivate idle nuclear reactors to meet the infrastructure and energy needs for AI research and development. Additionally, the Blueprint outlined a North American AI Alliance specifically to compete with foreign rivals.

02

Regulatory easing, greenlighting AI

The level of policy support and the extent of assistance provided to the AI industry have always been contentious issues during U.S. presidential elections. On one hand, various companies and politicians have continuously expressed concerns that the U.S.'s current stringent review system and tight electricity supply have severely hindered the progress of the AI industry. On the other hand, there is a widespread belief that Trump's administration will further loosen restrictions on the tech industry. For instance, Trump has announced plans to repeal President Biden's executive order on AI, claiming it 'impedes AI innovation and imposes radical left-wing ideas on the technology's development.'

Michelle Bowman, a governor of the Federal Reserve, acknowledged the risks posed by AI but warned that regulators should be cautious not to stifle the development of useful technologies.

In her latest speech, she stated, 'We do not need to rush to regulate. Overly conservative regulatory approaches may drive activity outside the regulated banking system or completely prevent the use of AI, thereby distorting the competitive landscape.'

This marks a significant shift in the U.S. approach to AI development and deployment. Under Trump's new administration, AI development in the U.S. will prioritize minimal regulatory oversight to facilitate AI progress and accelerate technological advancements through free-market methods. Trump's new approach emphasizes the principle that 'AI development is rooted in freedom of speech,' adopting a more laissez-faire regulatory style. On one hand, this approach caters to the broader tech industry, potentially accelerating AI development cycles. On the other hand, reduced regulatory oversight on AI safety and ethics may raise concerns about security and privacy standards.

The shift in AI policy under the Trump administration is heavily influenced by Elon Musk. On November 12, 2024, Musk officially took office at the U.S. Department of Government Efficiency. In the future, this key figure in the tech industry will also lead the reorganization of government operations, shaping how AI integrates into various government departments, becoming a central figure in the development and regulation of AI in the U.S.

The new regulatory environment can accelerate the development cycle and increase innovation in AI technology while speeding up industry consolidation in the AI sector. Large companies may have more freedom to expand their AI operations through acquisitions and collaborations, potentially reshaping the industry's competitive landscape. In particular, companies with interests aligned with Musk's may benefit significantly. For example, by relaxing AI regulations to allow fully autonomous vehicle deployment, reports suggest that the National Highway Traffic Safety Administration (NHTSA) may issue rules to simplify deployment. Currently, the U.S. only allows manufacturers to deploy 2,500 autonomous vehicles annually, but new legislation could raise this cap to 100,000. Tesla plans to produce driverless taxis starting in 2026, and if new regulations enable fully autonomous vehicles, Tesla CEO Elon Musk will directly benefit. Additionally, Musk's AI startup, xAI, is close to completing a new round of funding, with its valuation expected to double to $50 billion.

03

Other countries are also aggressively promoting AI, with rapid expansion of AI applications

In November, OpenAI announced the latest AI search experience for ChatGPT. Paid subscribers (and users on the SearchGPT waitlist) will gain access to real-time, internet-connected conversational capabilities, with free, enterprise, and education users gaining access in the coming weeks. Microsoft aims to create an ecosystem of AI agents, with key application scenarios already taking shape. On November 19, at the 'Microsoft Ignite 2024' global developer conference, Microsoft unveiled several updated AI products focused on AI assistants (AI Agents).

Furthermore, the South Korean government has proposed developing an AI-driven military to secure a leading position on future battlefields. The South Korean Defense Ministry and military also view AI technology as a crucial force multiplier, aiming to modernize the armed forces and effectively address future challenges. Additionally, SKT, a South Korean telecom operator, has successfully implemented its self-developed large AI model for telecommunications.

CITIC Securities released a research report stating that in Q3 2024, the deployment of AI applications accelerated further. Besides previously popular applications like Perplexity, which continued to grow rapidly, AppLovin and Palantir also demonstrated the productivity transformations brought about by AI. Simultaneously, a new wave of popular applications emerged, such as Hailuo, Kuaishou Keling, and Pika. Encouragingly, outstanding domestic AI applications began to stand out. With continuous technological advancements, more innovative products will emerge, particularly in AI marketing, AI imaging, AI virtual companionship, and AI military applications. Competition in these niche markets will intensify. As user demand for AI tools increases, we may see more customized solutions tailored to specific industry needs, driving further development across the industry.

With the acceleration of AI application deployment and the continuous emergence of real-world scenarios, coupled with the dual drive of cloud applications and terminal intelligence for industrial upgrading, companies in the AI sector are facing significant development opportunities. AI application scenarios are experiencing comprehensive breakthroughs and innovations, with cloud applications and terminal intelligence jointly driving industrial upgrading.

04

NVIDIA chips in high demand, capacity constraints

Currently, the U.S. leads China in computing power. Experts assess that one of the factors contributing to the dominance of U.S. companies (including OpenAI, Google, and Meta) in the global AI field is their preferential access to computing resources. Notably, these three companies are building computing infrastructure using tens of thousands or even hundreds of thousands of advanced NVIDIA GPUs (including the state-of-the-art GH100).

This year, NVIDIA became the world's most valuable publicly traded company, with its AI chips and a handful of anonymous customers propelling its market value skyward. Financial filings reveal that in the three months ending in October, Customers A, B, and C each accounted for 12% of NVIDIA's quarterly revenue. This suggests that these three customers likely received the maximum number of chips allocated to them by NVIDIA rather than receiving as many chips as they desired.

NVIDIA is unable to increase chip production as it has outsourced the wholesale manufacturing of its industry-leading AI microchips to TSMC, lacking its own production facilities.

Jensen Huang stated, 'We are in the early stages of ramping up production, which always comes with opportunities for volume increases. We are scaling Blackwell's production from zero to a very large level. By definition, the laws of physics dictate that the rate of volume increase is limited.'

The future deployment and development of AI applications depend on computing power and, more crucially, the production capacity and supply of AI chips. As the U.S. goes all in on the AI track, the supply of AI chips is paramount. This might explain why the U.S. is urging TSMC to set up a factory in the U.S.