Nvidia

Institutionalising luck at the world's 12th largest company by market cap.

Much of the information in this newsletter came from the Acquired podcasts discussing Nvidia.

One of the tricks to successful business is finding new use cases for your product or service. Who knew that you can turn breakfast cereal into a cheesy, something or rather? And before there was Panko, there were Cornflakes (see also, Cheddar Broccoli Corn Bake). Nike began as a company making running shoes for runners in America. This process of broadening the initial use case and market for a product sustains growth. The graphics chipmaker Nvidia is a great example.

On its most recent earnings call, Nvidia recorded yoy revenue growth of 46%. An astonishing achievement for a company that is 30 years old, with a revenue base of over $20bn. Growth comes from an ability to evolve, driven by its CEO Jenson Huang (not a dialectical gaussian), and its customers.

Customer Driven Evolution

From its beginning, Nvidia has been led by a willingness to invest and take risks, and an ability to be shaped by the consumer. Three episodes bear this out.

First, Nvidia began creating 3-D gaming chips for a market that did not exist. There were no developers ready for the product. But with other chip-makers, Nvidia built what became known as Graphics Processing Units (GPUs) and 3-D gaming was created (I spent too many hours at university playing Doom).

Second, Jenson received a letter from a professor. He’d had a theory but had been unable to run the calculations required to prove it. His gamer son suggested using Nvidia GPUs instead of normal CPUs. A GPU is programmed to calculate in parallel rather than sequentially, like a CPU. The professor bought the GPUs from a big box retailer, ran the calculation, and proved his life’s work. From this letter, Nvidia developed the CUDA program that has, after substantial investment, paid off in the last three years (see Data Centre in the chart below, 54% annualised growth).

Third, crytpo. In 2016 crypto miners identified Nvidia GPUs as the most efficient chips for mining bitcoin. Suddenly, crypto miners were buying Nvidia GPUs wherever they could be found, surprising the company, the market, and gamers, who could no longer afford their GPUs. Subsequently, Nvidia has formally addressed the market, built dedicated mining chips, CMPs, and made it impossible for crypto miners to use gaming GPUs.

These “random” episodes created rapid growth across segments, aside from crypto, for the moment (accounted for in Other).

Fact Interlude

For training speech recognition (speech to text, for instance), the number of mathematical problems you’re solving is greater than the number of grains of sand on earth.

The graphics card pictured below that fits in the palm of your hand has ~12 miles of lines drawn on it. Another reason to own ASML.

How does Nvidia grow in the future? Institutionalise Luck

Nvidia’s long-term growth rate from 2016, when the confluence of trends accelerated, has been 34%. The challenge for investors is to understand or, more likely, imagine how this streak can continue. One technique, used by Jenson himself, is to think in Total Addressable Market’s (TAMs). For instance, how big will self-driving cars be? Nvidia has the software platform. Lessep IM, instead, looks at understanding how growth mushrooms. This is the flywheel of software revenue growth; price, old user adds, new users, new use cases.

For Nvidia, they need to, and are, institutionalising luck through understanding how customers use their products. Over time, the instances when they are surprised by a letter from a client, or by crypto demand should decline as data identifies new and different ways products are used. If the assumption that the world is enjoying a productivity golden age is true, the current uses for Nvidia chips are understated.

Profitability

Nvidia, through the power of out-sourcing manufacturing to TSMC, has managed to look increasingly like a software company. Its gross margins of 65% are within range of a software company, Microsoft, for instance. As it continues to increase in scale, and maintains a software type model, Nvidia’s gross margins can improve further.

Of interest to software companies, Nvidia is a mature company with a strong Net Income Margin of 36%. The assumption in the Lessep IM software model is that, ultimately, 50% of the gross margin will end up as net income margin. But based on Nvidia, this number could be higher, and again increase software valuations.

Valuation

This allows a valuation exercise similar to that done for software companies. Nvidia must maintain a growth rate above 20% over the next five years, to be of value, but that is readily achievable based on our analysis of a company that can institutionalise luck.