Nvidia (US: NVDA) designs graphics processing units (GPUs) that underpin nearly all of our most exciting computing developments, from artificial intelligence (AI) to data centers and video games. Wherever there has been hype in the computing industry in recent years, GPUs produced by Nvidia or its rival AMD (US: AMD) are involved.
- Products Support Data Center Growth
- Many strong growth trends
- GPU market dominance
- Light capital, very cash-generating
- Slowdown in the games market
- Rich rating despite geopolitical tensions
Sitting at the intersection of several breakthrough technologies, Nvidia’s stock price has been particularly volatile over the past two years. In 2021, low interest rates coupled with a tailwind from the Covid-19 lockdown on its gaming business led investors to rise the stock 130%, to a peak valuation of 71 times forward earnings.
In 2022, it was a different story. Rising interest rates, tough comparators for the gaming industry and geopolitical uncertainty drove the stock price down 47%. Nvidia is now trading on a healthier price-to-earnings ratio of 40.
With inflation still at multi-decade highs and rising discount rates, this is still considered very costly. But like many tech stocks, it’s significantly cheaper than this time last year. Given Nvidia’s explosive growth potential in the coming years, this is also an entry point to consider.
Data drives demand
Before the turn of the millennium, computers ran on central processing units (CPUs) which were first commercialized by Intel in 1971. Although the power of CPUs improved rapidly in the first decades after their adoption, by the end of the century, this progress had begun to slow. Then, in 1999, Nvidia invented the GPU.
GPUs allow for faster computation because they use matrix computations rather than linear computations. This is called parallel computing. To process so much information, GPUs need hundreds of “cores” as opposed to the handful that sits on CPUs. CPUs are still used in computers for simpler tasks, but nowadays all the really heavy-duty tasks like graphics generation, machine learning, and autonomous driving are done by GPUs.
Nvidia started by designing GPUs for PCs to improve gaming graphics, but later expanded into data centers, professional visualization, and the automotive industry. The company is fabless, meaning it outsources production, mostly to the semiconductor giant TSMC (TW:2330) In Taiwan. While this model leaves the company dependent on third parties, it also makes Nvidia a capital-light, high-margin, and extremely cash-generating company. In 2021, the operating margin was 40% and free cash flow reached $8.13 billion (£6.8 billion) from just $10.7 billion in earnings. exploitation.
In the third quarter of this year, data centers accounted for 65% of group sales. Games accounted for another 26%, followed by automotive and professional viewing at around 4% and 3%, respectively.
Despite this, data centers have only just become the company’s dominant division. Third-quarter data center revenue grew 31% year-over-year to $3.83 billion, driven in part by Amazon Web Services’ growing use of the NVIDIA A100 Tensor processor Core in its servers. Nvidia also announced new two-year partnerships with Oracle (US: ORCL) and Microsoft (US: MSFT) during the quarter, the latter covering a contract to “build an advanced cloud-based AI supercomputer to help enterprises train, deploy and scale AI.”
Almost all enterprise software now runs on cloud servers rather than on premises. The cost of running these servers is so high that it doesn’t make economic sense for a midsize business not to outsource. In Amazon’s last quarter, Alphabet (US: GOOGL) and Microsoft all grew their cloud business by more than 30% at constant currency compared to last year. As demand for cloud services increases, Nvidia’s revenue will also increase.
The advantage of investing in Nvidia as opposed to cloud computing companies themselves is that Nvidia has little competition in the GPU market. Intel (US: INTC) and AMD make processors for cloud servers, but neither can yet challenge Nvidia in designing GPUs used for machine learning and AI. It was only last year that Intel launched its first GPU for data centers.
An AI inflection point
Until now, AI has been almost exclusively used by companies for tasks such as improving cybersecurity or optimizing customer data. However, pathways to more mainstream applications of AI have emerged from text-to-image generators like DALL-E 2 and the GPT-3 language model. DALL-E 2 can turn text prompts into eerily precise images. It does this by scraping all the images from the internet, a task that requires so much computing power that it has to be run in the cloud rather than on a laptop. Last month, Microsoft’s cloud computing division, Azure, said its customers would be able to run DALL-E 2 through its Open AI service.
Although Microsoft only provides these services to enterprise customers, Nvidia says consumer AI applications — such as “big language models, recommender systems, and generative AI” — are already driving growth. alongside the main cloud players.
This is an encouraging development. While new technologies are often first adopted by enterprises, demand often kicks in when consumer applications emerge. It happened with the personal computer and now it’s starting to happen with machine learning. It seems to have been a turning point.
A difficult context
While the data center business is getting better and better, gaming has been a weakness. Game sales in the three months to September were down 51% year-over-year. Indeed, Nvidia’s partners needed to “align channel inventory levels with current demand expectations as macroeconomic conditions begin to weigh on consumer demand.” In other words, when the economy is bad, people spend less money on expensive gaming laptops.
Gaming GPUs are also used for cryptocurrency mining which boosted sales this time last year but is now down. This translated into a 31% drop in gaming revenue over two years, during which time data center sales doubled.
The consumer-facing arm was still more likely to be affected by the downturn, especially after the lockdown-induced boom. But that doesn’t mean it won’t rebound, as record opening weekend sales for the last Call of Duty the title suggests.
Other trends may be more difficult to untangle. Nvidia’s strong data center numbers in the third quarter landed despite the recently passed US Chip and Science Act, which limits the export of certain chips to China that can be used for supercomputing and AI. For Nvidia, this means there will be no Shenzhen-bound sales of its A100 and H100-based products.
It’s unclear how much this will affect the company. During its third-quarter results, Nvidia said “sequential growth was impacted by weakness in China,” but did not specifically tie that to legislation. However, any form of further escalation between the United States and China over Taiwan would be worrisome for investors. Last year, 26% of revenue came from China, 16% from the United States and 32% from Taiwan.
To circumvent the sanctions, Nvidia has already designed a less powerful GPU that can be sold to Chinese data center companies but cannot be used in supercomputing. However, there is always the possibility that the United States will change the law again to prevent the shipment of any GPU designed in the United States.
Knowing what discount to apply to this geopolitical uncertainty is tricky. But it certainly contributes to a multi-year discount to the market valuation of a company with near-infinite potential.
|Company Details||Last name||Market cap||Price||52 weeks Hi/No|
|Nvidia (NVDA)||$394 billion||$158.27||33,412¢ / 10,813¢|
|Size/debt||NAV per share*||Net Cash / Debt(-)*||Net debt/Ebitda||On Cash/Ebitda|
|Evaluation||PE before (+12 months)||DD (+12 months)||FCF yield (+12 months)||GV^ ratio|
|Quality/ Growth||EBIT margin||YEARS||CAGR of sales over 5 years||CAGR EPS 5 years|
|Forecast / Momentum||Fwd EPS grth NTM||Fwd EPS grth STM||Mom of 3 months||% change in EPS before over 3 months|
|Year-end Jan 31||Sales (in billions of dollars)||Profit before taxes (in billions of dollars)||EPS (¢)||DPS (¢)|
|To change (%)||+9||+36||+31||+37|
|Source: FactSet, adjusted PTP and EPS figures. NTM = next 12 months. STM = Second 12 months (i.e. in one year). * Includes intangible assets of $6.7 billion, or 269¢ per share. ^GV Ratio = (EV / Ebit) / (Fwd EPS grth + DY)|
#Nvidia #worth #hefty #price #tag