Governments are securing graphics processing units (GPUs) for AI development; UK and Saudi Arabia investing heavily. Saudi Arabia and the United Arab Emirates have purchased thousands of Nvidia’s H100 chips for AI research and innovation, despite the high cost of $40,000 per chip. The UK government has committed £100m to order components from Nvidia, AMD and Intel to improve the country’s AI capability.
Governments around the world are struggling to secure the supply of graphics processing units (GPUs) needed for AI development and processing.
With the most sought-after chips in short supply, countries like the UK and Saudi Arabia are spending significant financial and diplomatic resources to secure enough GPUs for their domestic AI sectors.
Investors have pointed to Nvidia as one of the potential big winners in AI, following the viral success of the ChatGPT chatbot and other popular tools. The company’s chips excel at parallel processing, making them well-suited for training software by bombarding it with data.
Middle Eastern Governments Spend Big on Nvidia Chips
According to a report this week in the Financial Times, Saudi Arabia has bought at least 3,000 Nvidia H100 chips for King Abdullah University of Science and Technology with the intention of building a new supercomputer. The report indicates that the UAE has also invested in thousands of Nvidia chips through state-owned companies.
In those countries, governments are investing heavily to help advance AI research and foster innovation. And they need to. At $40,000 each, the H100s don’t come cheap. But they have become almost essential in the world of advanced AI development.
Few technology companies have the resources to train large AI models like OpenAI’s GPT-4. Due to their high processing power, expensive Nvidia chips are highly prized for these kinds of tasks. The GPU maker has become a key player in the contemporary rise of AI. Yet even so, OpenAI’s supercomputer contains 10,000 A100s, a predecessor of the H100.
UK AI Sector to Boost GPU Supply
In a bid to boost the UK’s AI capacity, it was reported on Sunday 20 August that Prime Minister Rishi Sunak has committed £100m of public money to order key components from Nvidia, AMD and Intel. The report notes that the government is also in advanced stages of obtaining GPUs from Nvidia.
However, in the long term, the challenge of maintaining a technological advantage is much more complex and dynamic. For countries like the UK, stockpiling GPUs is only a temporary solution to current challenges.
National Manufacturing Capacity Creation
Part of the reason graphics cards are so expensive is that the entire GPU supply chain is constrained by the manufacturing capacity of chipmakers in Taiwan. Commenting on the tensions between Taiwan and China, the Nvidia CEO said he feels “perfectly safe” in being so reliant on supplies of chip power.
Governments around the world, though, have flagged the problem as a national security concern. And many see bolstering their domestic manufacturing capacity as key to maintaining a secure supply of chips.
The UK government, for example, has published a national semiconductor strategy. Over the next 10 years, the plan calls for a billion-pound investment in the country’s chipmaking sector.
However, that number is dwarfed by subsidies in the EU and the United States, which have committed $52bn and £43m respectively to support semiconductor manufacturing.
By Audy Castaneda
