The government is potentially upsetting the AI rally in US equity market as it is considering expanding the export controls put in place back in October 2022 on AI chips to China. Nvidia shares are down 4.5% in pre-market trading and it could quickly spread to other US technology stocks. Nvidia 's geographical revenue figures show that China was by far the biggest driver of higher revenue in FY24 Q1 (ending 30 April 2023) as Chinese technology companies are scrambling to catch up to US technology companies in AI technology. Expanded export controls on AI chips are by far the biggest risks to Nvidia 's exuberant outlook provided back in May, so investors should consider their exposure at this point.
- The US government is considering expanding the export controls on AI chips related to China as the US government aims to limit China’s ability to catch up to the US on AI technology and use advanced AI chips in military applications.Yesterday, after US equity markets closed, the Wall Street Journal announced in an exclusive article that the US government is considering adding new curbs on exports of AI chips that would most likely require Nvidia to get a license requirement for specific customers in China. The timeframe mentioned in the article in which the US Commerce Department would make a decision is already early July. This expedited timeline suggests that the US is fearing that China is doing everything it can to catch up to the US lead in AI technology. Nvidia shares are down 4.5% in pre-market trading with a potential key breakdown of the 400 level in today’s trading session should investors reduce their exposure to Nvidia . AMD shares are down 3.5% in pre-market trading.
The potential new export restrictions are key for market sentiment as the AI-hyped rally in US equities has been a key driver of performance this year. Nvidia ’s strong outlook presented in its FY24 Q1 results (ending 30 April) on 24 May has lifted the entire cluster of AI-related stocks and turbocharged 12-month earnings expectations for the S&P 500 technology sector ahead of the important Q2 earnings season starting in mid-July.
To understand the importance of these potential new export restrictions facing Nvidia and AMD we have to go back to 7 October 2022 when the Biden administration issued controls on the export of advanced US AI and semiconductor technology to China. This policy was designed to prevent advanced chipmakers such as Nvidia exporting its A100 chips to China. Nvidia shares declined 12% over four trading sessions as investors realised that around 26% of the business was directly threatened by this new policy.
Already a month later news surfaced that Nvidia was working on a solution so that it could export AI chips to China that would be within the limits of the new rules. The codename was A800 and based on recent 10-K filings the company mentions A800 and H800 chips for the Chinese market as replacements for the successful AI chips A100 and H100. The change that Nvidia did was to lower what is called the high-speed interconnection bus known as NVLink to maximum of 400 GB/s. The change opened up the Chinese market for Nvidia ’s new AI chips without a license requirement. As this article highlights there was a strong uptick in Chinese demand for A800 and H800 already in April driven by purchases from Tencent and similar technology companies in China. This aligns well with the facts that China is significantly behind US companies on AI and that the launch of ChatGPT and Bard has ignited a sense of urgency in China to catch up with the US. Chinese media has described the Chinese government’s intentions to increase research and development significantly this year and a Bloomberg article today is highlighting that up to $15bn is expected to flow into AI spending this year alone in China.
If we look closer at the geographical revenue figures for Nvidia we see the explosion in demand from China as well. Nvidia never mentions neither cryptomining or China in its press releases or investor presentation slide decks. This is intentional. To find the geographical revenue figures one must dig deep into the 10-K filings. Based on those filings we see a clear pattern. In 2022, revenue declined in China and Taiwan as cryptomining became significantly less profitable due to collapsing cryptocurrencies including Bitcoin . The market got flooded with Nvidia ’s GPUs as they were no longer in use for cryptomining.
Another striking observation is the rapid decline in Chinese revenue in the quarter than ended on 29 January 2023 despite cryptocurrencies stabilised. Our assumption is that this decline was because of the new export restrictions introduced on 7 October 2022. What we also see in the same quarter is a jump in other countries. We know from the recent 10-K filing that Singapore, which has now been split own separately, is the biggest country in that segment and thus Singapore could have been used as an intermediary for the A100 and H100 chips.
In FY24 Q1 that ended 30 April 2023, when Nvidia put out its exuberant outlook, we see that Chinese revenue increases 67% q/q while the US market is only up 7% q/q despite this is the AI powerhouse. The other countries segment is down 31% q/q. It is quite striking that despite a hefty increase in Chinese revenue, Nvidia is only mentioning China once across their press release and investor presentation and that is a small note under its Automotive segment, that “some NEV customers in China are adjusting their production schedules to reflect slower-than-expected demand growth”.
If we put all the data points together it points towards that the new A800 and H800 are driving a lot of demand in China as Chinese technology companies are scrambling under directions of the government to catch up with US technology companies in AI. The fact that the US Commerce Department is considering changing the export rules suggest that the US government has picked up on this trend as well and is getting worried about what it means for the US leadership in AI technology.
The risk of new export restriction rules is by far the biggest risk to Nvidia shareholders as license requirements for AI-related chips would significantly impact financial results in a negative way as described in Nvidia ’s 10-K filing. In the event that new US government export controls are put in place on AI chips this could deflate the AI rally as it would set off risk reductions across the entire US technology sector as funds would likely reduce exposure and take profit.
As we described in our Q2 Quarterly Outlook, the world has entered into the fragmentation game which is an extension of the trade war that the Trump administration started in 2016. The fragmentation game is a geopolitical dynamic centred around the idea of derisking global supply chains. In the beginning Europe was trying to play a neutral game but Russia’s invasion of Ukraine pushed Europe into the fragmentation game full throttle. The fragmentation game has four main pillars and those are 1) defence, 2) energy, 3) commodities, and 4) technology.
If AI technology is the most important technology vector over the next decade there is no chance that Nvidia will be able to play the neutral commercial playbook maximizing shareholder profit. The company will be forced to strictly adhere to US industrial and national security policies related to AI technology. Given that expectations are currently being formed by Nvidia ’s recent outlook which predominantly driven by Chinese technology companies being forced to play catch up to US technology companies there is a serious risk to Nvidia ’s outlook should the US government go along with new export controls on AI chips to China.Investors sitting on strong gains in Nvidia and US technology companies in general should consider de-risking their portfolios by lowering their exposure. We are seeing peak expectations and those expectations could be crushed under the fragmentation game. A negative impact on Nvidia will likely spread to the “AI risk cluster” which consists of stocks such as AMD, ASML , Adobe , Alphabet , Meta, Advantest, Palantir, Marvell Technology , Microsoft , and Applied Materials .