Saudi Arabia and UAE race to buy Nvidia chips to power AI ambitions

[tonyget]

Saudi Arabia and UAE race to buy Nvidia chips to power AI ambitions | Financial Times

Gulf nations acquire thousands of GPUs amid global shortage of semiconductors needed to build large language models

www.ft.com

Saudi Arabia and the United Arab Emirates are buying up thousands of the high-performance Nvidia chips crucial for building artificial intelligence software, joining a global AI arms race that is squeezing the supply of Silicon Valley’s hottest commodity.

The Gulf powerhouses have publicly stated their goal of becoming leaders in AI as they pursue ambitious plans to turbocharge their economies. But the push has also raised concerns about potential misuse of the technology by the oil-rich states’ autocratic leaders.

According to people familiar with the moves, Saudi Arabia has bought at least 3,000 of Nvidia’s H100 chips — a $40,000 processor described by Nvidia chief Jensen Huang as “the world’s first computer [chip] designed for generative AI” — via the public research institution King Abdullah University of Science and Technology (Kaust).

Meanwhile, the UAE has also secured access to thousands of Nvidia chips and has already developed its own open-source large language model, known as Falcon, at the state-owned Technology Innovation Institute in Masdar City, Abu Dhabi.

“The UAE has made a decision that it wants to . . . own and control its own computational power and talent, have their own platforms and not be dependent on the Chinese or the Americans,” said a person familiar with Abu Dhabi’s thinking.

“Importantly, they have the capital to do it, and they have the energy resources to do that and are attracting the best global talent as well,” the person added.

The Gulf states’ purchases of large numbers of Nvidia chips via state-owned groups come as the world’s leading tech companies rush to obtain the scarce chips for AI development.

Last week the Financial Times reported that leading Chinese tech groups such as Tencent and Alibaba are seeking to buy up Nvidia’s high-performance chips.

The most advanced LLMs are owned by US companies including Microsoft-backed OpenAI and Google. These groups are also the primary buyers of Nvidia’s H100 and A100 chips.

According to multiple sources close to Nvidia and its manufacturer, Taiwan Semiconductor Manufacturing Company, the chipmaker will ship about 550,000 of its latest H100 chips globally in 2023, primarily to US tech companies. Nvidia declined to comment.

In Saudi Arabia, Kaust will receive 3,000 of these specialist chips, worth about $120mn in total, by the end of 2023, according to two people close to the university’s AI labs.

By comparison, estimates suggest OpenAI trained its advanced GPT-3 model on 1,024 A100 chips — the predecessor to Nvidia’s latest chips — in just over a month.

The Saudi university, which, according to people close to Kaust, also owns at least 200 A100s, is building a supercomputer, Shaheen III, that will become operational this year. The machine will run 700 Grace Hoppers, Nvidia’s so-called superchips, designed for cutting-edge artificial intelligence applications.

Kaust will use these chips to build its own large language model — software that can generate humanlike text, images and code — similar to OpenAI’s GPT-4, which powers the popular chatbot ChatGPT, according to multiple sources close to the state-owned university.

The Saudi LLM is being developed by the Provable Responsible AI and Data Analytics lab at Kaust, which is primarily staffed by Chinese researchers.

Many Chinese nationals with AI expertise have chosen to work at Kaust because they have been prevented from studying and working in the US after graduating from Chinese universities on the US entity list, according to two Kaust sources.

Kaust did not respond immediately to requests for comment.

Meanwhile, the UAE, which in 2017 was the first nation to establish an AI ministry, has launched a “Generative AI Guide” as part of the government’s “commitment to reinforcing its global position as a pioneer in technology and artificial intelligence sectors” as well as “regulatory frameworks to limit the negative use of technology”.

The UAE’s Falcon model, which is now freely available online, was trained on 384 A100 chips over two months earlier this year.

“I was extremely impressed by the model, considering the resources they used. For a while, it was among the best models in the open-source world,” said one leading AI researcher and LLM expert.

It also impressed well-known venture capitalists such as Marc Andreessen, who tried to establish contact with the team, according to a UAE sovereign wealth investor. A spokesperson for Andreessen declined to comment.

The UAE government has purchased a new batch of Nvidia chips to prepare for more LLM-related applications and cloud services, according to a person at the UAE industrial development bureau.

Western AI leaders and human rights experts have raised concerns that the software developed in the two countries may lack the ethical guardrails and safety features that large tech companies are trying to build into the technology.

“Human rights defenders and journalists are frequent targets of government crackdowns [in UAE and Saudi Arabia],” said Iverna McGowan, director of the Center for Democracy and Technology’s Europe office, in Brussels.

“Pair this with the fact that we know how AI can have discriminatory impacts, or be used to turbocharge unlawful surveillance. It’s a frightening thought,” she added.

Saudi Arabia and the UAE have been buoyed by petrodollar windfalls after last year’s surge in energy prices, and both manage some of the world’s largest and most active sovereign investment funds.

Representatives of Gulf state-affiliated funds have recently approached AI start-ups in the west in an effort to access code and LLM expertise in exchange for computing resources, according to the chief executives of two European AI companies, who declined their offers.

“We have had offers of immense financings and access to data, in order to tap indirectly into our talent,” one of the executives said.

OpenAI chief executive Sam Altman in June praised Abu Dhabi’s foresight in recognising the importance of AI during a visit to the region. In a question-and-answer session in the city’s financial district, he said the Gulf region could “play a central role in this global conversation” around the emerging technology and its regulation.

“There has been discussion about AI, here in particular, in Abu Dhabi, before it was cool,” he said. “Now everyone is on the AI bandwagon, which we are excited about, but we have special appreciation for the people who were talking about this when everyone thought AI was not going to happen.”