Nvidia's post-Q4 Performance and U.S. Sanctions on AI Tech Exports
Quick summary of recent media coverage and commentary about the year's hottest stock (so far).
Over the past couple of weeks, I handled a number of media inquiries about Nvidia’s Q4/FY 2023 results. Subsequently, my commentary was highlighted in Investing.com with the title “This analyst says Nvidia's stock "‘appears vulnerable to a correction’”, along with mentions in Börse München (in German) as well as in MarketBeat. Here’s a quick recap. Read on!
As outlined in my article on Leverage Shares as well as SeekingAlpha, Nvidia’s consumer mix has decidedly shifted towards a “corporate” client base over the past couple of years. Among its segments:
The largest revenue contributor was the data center segment, which registered a 217% year-on-year change at $47.5 billion, with numerous enterprise solutions based around language models, generative AI and medical applications.
The 2nd-largest year-on-year change (21%) and the smallest revenue contributor ($1.1 billion) was the automotive segment, with increased adoption by Chinese carmakers Great Wall Motors and Li Auto lying square and center.
As the company’s consumer mix changed over 2023, so did the company’s Price-to-Earnings (PE) Ratio:
As of the day before the earnings release, the PS Ratio was down 38% from a high of 45.5 while the PE Ratio was down 74% from a high of 222. After the earnings release, the PE Ratio shot up to the 85-95 range.
In a pool of 120 semiconductor/chipmaking-related stocks (in which Nvidia is included), investors were fairly measured in their outlook up until Q4 2022.
Despite Q1 2023 being a relatively bad period for stock performance, investor vigour contributed to a doubling of PE Ratios over the previous quarter. Barring a slight adjustment upwards in the next two quarters, massive net income accruals in Q4 2023 pushed the sector’s PE aggregate to a twelvefold of its value a year prior. For Nvidia, it’s unlikely that revenue/net income accruals will be matched at this high a level in the quarters to come.
The company predominantly operates as a designer while the bulk of the manufacturing/assembly is carried out by Taiwan-based TSMC. The company’s “high water mark” currently achieved via successful corporate outreach and collaborations comes with the caveat that its counterparts will constantly look towards optimizing cost and performance based on need. Thus, there are two pressure points: one from the manufacturer and the other from its clients/other counterparts.
Thus, a repeat of historically high ratio premia being sustained cannot be expected. A further 40-50% correction of the PE Ratio from the present level to the mid-twenties to early-thirties is likely on the cards, which will be somewhat complicated by the fact that the board authorized an additional $25 billion in share repurchases “without expiration” during the Q2 earnings release. Strong buy volumes would lift the stock price up momentarily but this should be read as a “false positive” on overall trajectory.
Conversations were also had with reporters from Reuters and Automotive World about Nvidia’s business in China versus the U.S. sanctions regime regarding AI-relevant technology. Here’s the background behind my opinions. If/when they get mentioned, the articles will be linked here.
Nvidia’s relationship with China is particularly important. Nvidia’s record-breaking earnings came amidst ongoing of U.S. Commerce Department restrictions on exporting advanced chips and chipmaking equipment to China first introduced in October 2022, which knocked the company’s A100 and H100 chips out of China. The company’s “China-tailored” A800 and H800 chips were limited in restriction amendments in October 2023, following which the company announced another set of chips — the H20, L20, and L2 GPUs — that it announced would be compliant with restrictions. However, early reports indicate that this isn’t necessarily true, with a number of wide-reaching industry commentators stating that at least one of these was 20% faster than the proscribed H100 in some respects.
Within its blockbuster data center segment, China accounts for nearly 25-30% of Nvidia’s revenue as the likes of Baidu, Alibaba and Tencent accelerated spending on AI integration. In its ascendant automotive segment, it’s almost entirely dependent on Chinese automakers focused on using AI-relevant technology to deliver “enhanced” driving functionality in their products. China’s carmakers have been logging up some advances in this front: for instance, Baidu’s “Apollo Go” autonomous driving division had — as of the end of 2023 — accumulated over 5 million rides, with fully driverless orders in Wuhan (a city of some world renown since 2020) being nearly half of all orders.
Nvidia’s DRIVE™ Thor, its next-generation superchip, is virtually designed for autonomous vehicles with 2,000 teraflops of performance and intelligent functions packaged into a single architecture for maximum efficiency. The argued forward-looking value of automated vehicle cannot be overstated: while initial costs may be high, the absence of a human driver theoretically ameliorates operational costs for large-scale fleet operators in commercial applications.
But Nvidia’s not alone in monetizing this space: both Intel and Qualcomm are now entering with products of their own — with the latter scoring some early traction by inking deals with BMW and Mercedes-Benz. Intel, which already operates in the autonomous driving space via subsidiary MobileEye, estimates that increasing intelligent automotive functioning would be a net positive for chipmakers’ wallet share in new automobiles: processors in 2019 represented only 4% of a car's cost, but it's going to soar 500X to 20% by 2030.
So, the first reason as to why American chipmaker Nvidia is seemingly adhering to the “letter of the law” from the U.S. administration as opposed to the “spirit of the law” with regard to supplying China with AI-relevant hardware is that it will be a significant impact on the company’s bottom line both in the present and in the forward-looking outlook. The second reason might be more pragmatic.
Nvidia isn’t the only company that’s dependent on an “external” company for the manufacturing of the products it sells to its customers. TSMC accounts for 90% of the entire world’s demand for chips rated “10 nanometers or below”, i.e. the chips needed for highly advanced applications. While China is working on boosting domestic production, it is expected that the technology gap will only continue to grow as years roll by.
However, as Taiwan’s chipmaking foundries roll out greater volumes of product, the greater will be their dependence on China for raw material.
Overall, it is estimated that nearly half of all the parts needed by TSMC to manufacture chips will have to be sourced from China by 2030.
If the U.S. administration continues to ramp up its “chip sanctions”, there will likely be a period of forbearance by China given that upwards of 90% of all semiconductors it uses are imported or manufactured locally by foreign companies. But, if push comes to shove, there is potential for China to choke the likes of Nvidia by restricting the export of parts to the foundries of Taiwan. This could be construed as the second reason for Nvidia treading lightly when it comes to Chinese demand.
The ostensible reason for the restrictions on AI-relevant technology to Chinese government and military entities. However, a review of tender documents by Reuters earlier in the year revealed that the “banned” chips are openly and plainly being acquired by those very entities via “unauthorized resellers” in India, Taiwan and Singapore. Given that these resellers aren’t beholden to the likes of Nvidia and that these chips are openly available throughout the world, these “sanctions” (arguably) place unreasonable demands on the likes of Nvidia. As the complexity of the chips on offer by Intel and Qualcomm increase, it can be expected that they too will have to contend with compliance with these (arguably ineffectual) sanctions.
The “dual use” of technology is almost as historied as humanity. How the U.S. administration applies restrictions in an era of unprecedented global trade is a question it must ponder upon.
The “Dharma” series — which traces the evolution of Eastern faith and philosophy — continues to draw attention in the year so far. Here’s Part 1, Part 2, Part 3, Part 4 and Part 5, followed by the ancillary Part 6 and Part 7 discussing Malaysia’s and Indonesia’s spiritual history.
Also, click here for a list of all articles ever published.