By Ganesh Natarajan
The world has been watching closely the developments in artificial intelligence (AI) in the US and China. Most of the noise has emanated from Nvidia, OpenAI, Microsoft, and Google, and the astounding valuations enjoyed by Nvidia and OpenAI. Nvidia’s production of graphics processing units (GPUs) has created such demand that the valuation of the company crossed $3.2 trillion. ChatGPT and, more recently, the video-to-text model Sora from the OpenAI stables can take credit for setting off the new wave of interest in AI and taken the company to a valuation of over $150 billion in its last fundraising. Despite protests from climate change activists over the enormous energy-guzzling demands of GPUs and large language models (LLMs), applications have burgeoned in areas ranging from architecture, engineering, and construction to media and entertainment, automotive, scientific research, and manufacturing design. They have necessitated investments in edge-to-cloud computing and in supercomputers and workstations for applications in these fields.
In this context, it is not surprising that the hot news of the day is the Stargate project, a massive $500-billion joint venture announced by Donald Trump, with Oracle and OpenAI as key partners. This initiative aims to boost AI infrastructure in the US, creating over 100,000 jobs and securing pole position for America’s leadership in the contest. It has also been conceptualised with multiple partners. Nvidia and Microsoft will continue to be the key technology partners. Oracle will be a technology contributor and also a major infrastructure enabler. And OpenAI will lead the project’s operations with SoftBank overseeing the financial commitments of various parties.
With an initial investment of $100 billion already in place and plans to deploy the balance $400 billion over the next four years, one can expect 20 massive data centres, each spanning over 500,000 square feet, to come up. This will be a key enabler to the new approach to industrialisation in the US and allow the country to lead in the much-anticipated advent of artificial general intelligence (AGI), and mitigate some of the fears and uncertainties associated with a future involving artificial super intelligence (ASI).
Stargate epitomises everything that Trump 2.0 stands for — namely a large, collaborative US initiative which celebrates large-scale technology providers like Oracle, Microsoft, and OpenAI, and underlines the MAGA vision of America being the creator and leader of all that is good and great for mankind. The approach builds on the big investment leadership of Nvidia and the LLMs of OpenAI. It will give the US the dominant 500-pound gorilla status in high-investment, high energy-consuming AI infrastructure. It is surprising that IBM has not been invited to join the party. But there is no doubt that the early participants will be the big beneficiaries as already seen in the sharp jump in the stock price of Nvidia, which will be an early winner with the huge capacity creation in data centres.
Industry and AI watcher Sangeet Paul Choudary suggests that the new “sandwich” that is being created with Stargate will enable a strong base of infrastructure that cuts across industries and a top layer for market-facing technology aggregators like Meta, Amazon, and Google. This could create an economic framework for early players to develop and deploy AI, capture the resultant market value, and simply “change the rules of competition for everyone else”.
Stargate fits perfectly into the “large is beautiful” approach of the new administration and there is no doubt that the US can and will capture leadership in this “America first” approach to AI. But the world should watch out for alternative models of using AI for corporate and public good that will emerge from research labs in China and possibly Japan, Vietnam, Korea, and later on Israel and Saudi Arabia.
China will be shut out of Stargate Technologies, and the lack of access to Nvidia processors will and probably already has resulted in the country focusing on a smaller but equally effective approach through narrow language models. Point AI and Application AI solutions through small language models (SLMs) from China and possibly India could reinvent the future of AI.
At this point, the US is defining the rules of the game. A research report recently released by Stanford University’s Institute for Human-Centered AI positioned the US far ahead of China and all others primarily because of the large $60 billion dollar-investment by the private sector in AI as compared to just about a tenth of that in China. However, China continues to file patents at a rapid clip and invest in technologies for the future. Its research community has proved extremely capable and prolific, particularly in the universities. In India, the public discourse and resolve to be leaders in research and innovation in this critical field shows great promise. But as always, the rhetoric on social media needs to be backed up by strong government and private sector research to merit a seat at the high table of AI for years to come.
If we have to put climate change and sustainability back on the table, the world should explore nimble, cost-effective options to the Stargate approach. With the strong backing of Chinese and Indian researchers and possible innovations coming from the Gulf Cooperation Council and the Association of Southeast Asian Nations, we need a focused approach beyond chip manufacturing, process automation, and technology services for a true alternative to the might of Stargate. We are witnessing the beginnings of a revolution that is as significant as the wide proliferation of electricity in the world and the future of prediction, prescription, and dual intelligence where human advancements in innovation and technology advances in AGI and ASI will see truly transform business and society. May the AI games begin!
(The author is the Chairman of 5F World and GTT Data Solutions Ltd)
(Disclaimer: Views expressed are personal and do not reflect the official position or policy of Financial Express Online. Reproducing this content without permission is prohibited.)