1000X Soar: What Makes Cash Allocation So Successful?

This time Nvidia, finally. I know you say, Nvidia has leaped because of the AI trend, and the stock price went to the sky. So what? This company is so big, nothing applies to our company?

Since its IPO on January 22, 1999, the stock price has become 1000-fold from $0.8 to $900 in Apr 2024. Even before the AI boom in 2023 occurred, the stock price was more than 100 times higher than the IPO price.

But until 2015, the company was just a mid-sized normal Chip design company. The stock price around $5 hadn’t changed for 10 years since 2005. Intel’s market cap was 10 times bigger than Nvidia’s. Now, Nvidia is 10 times bigger than Intel.

I like Nvidia because it kept focusing on niches, not only on GPUs but narrower. A GPU is a PC chip; when you play a game like Final Fantasy with heavy graphics, it requires high computing power. So, that high computing power can be used for dealing with AI. They focused first on game graphics chips, then they shifted their focus to AI chips because they thought AI was the future of humanity. Apparently, to grow the AI industry, a GPU specialized in AI is necessary. Other chips can’t handle it.

Their USP (Unique Selling Proposition) is that it is really fast, the fastest chip when it comes to handling big data that AI uses. No other chips like AMD’s match the processing speed. If an AI company does not have Nvidia’s state-of-the-art A100, they are losing beforehand. Even ChatGPT, the most successful AI, is using A100, which fact is worth $1 Trillion. A100 is crazy expensive at $10K (thousand) compared to its GPU for game PCs, which normally costs less than $1K, because they can sell it at a very high price. The gross profit rate ((sales-cost of sales)/sales) for Nvidia’s AI chip is really high.

The chip is so efficient that no other chips can reach the level because Nvidia has been betting big on AI. The company spent big on AI chip Research and Development (R&D) from early times, way before the ChatGPT boom came. No other company’s total R&D spending on AI GPUs exceeds Nvidia’s. The company kept accumulating knowledge.

So in 2023, when the AI boom came, the company sold way more AI chips, and the gross profit rate became really high, from 57% to 73% (in Jan 2024) in a year. The company’s net income rate (net income/sales) became 49%, insanely high even though they spent $8.6B (Billion) on R&D. Their “rival” AMD (by the way, the market cap is now one-tenth the size of Nvidia)’s net profit rate was only 3.7% in 2023. AMD couldn’t focus on AI GPUs in the past, and they have to sell cheaper chips, so they have a lower gross profit (sales-cost of sales). But it has to spend on R&D to keep up with the AI chip competition. The R&D spending was $5.8B in 2023 for AMD, and that reduced AMD’s profit significantly. The R&D spending rate was really high for AMD but the absolute amount was less than Nvidia’s.

About AMD’s scattered focus, for example, you can’t find the word “AI” (you can find words like Dub”ai” but apparently not “artificial intelligence”) in their 2016 financial statement (FS). Now they are crazy about AI , but too late. Their main focus in 2016 was how to compete with Intel. AMD had to spread its focus to GPUs and CPUs (normal PC chips for PC operations), and their GPU spending was only on graphics purposes, not AI. Now AMD has 4 segments: 1. Data Center 2. Client 3. Gaming 4. Embedded segment. You can see how focusing on a niche area is important.

Great companies have to accumulate cash from efficient operations and invest it for future cash flow. Cash flow is very different from profit. Many companies are profitable but always short of cash.

That happens when their operations are not efficient, like when they have too much inventory or equipment. They have to keep putting additional cash into those operating assets, never getting the cash back. To understand this, suppose at the beginning you have $100 worth of chip inventory, you sell one for $200, but then you purchase two more for $200. At the end of the day, the profit is $200 – $100 = $100, but the cash is $0 ($200 – $200). This is a bad operation.

Nvidia too had to accumulate cash to invest in AI. Its main source of cash flow was always its thick gross margin. As I mentioned, Nvidia’s gross profit rate is very high. Even if the operating efficiency is normal, it still generates a lot of cash. For example, in the same situation above, but you sell the chip for $500, then the cash flow is $300 ($500 – $200).

Nvidia accomplishes this by focusing on niches, making products before the market is available, and becoming the first person to sell the product/service, accumulating knowledge, and making a strong brand recognition in the industry – first game graphics, then AI.

Nvidia vs Meta: What Separates Great Decisions from Bad Decisions

The difference between Meta’s reckless metaverse investment and Nvidia’s AI investment (which actually generates a crazy amount of profit, making Nvidia the 3rd biggest company in the world) is timing.

The metaverse is still not good enough technologically or in people’s acceptance. Making graphics still costs too much, and people are not impressed. I believe the metaverse has to wait until we can access online through brain chips directly because people hate headsets.

Also, people have no reason to go to the metaverse. It seems just a silly game. On the other hand, if people do not adopt AI, they lose competitiveness, so people and companies must adopt it.

Meta spends too much on the metaverse at too early a stage. Now is the metaverse winter. Meta spends money to teach future rivals what doesn’t work and also what doesn’t. It has experienced only failure.

On the contrary, Nvidia entered the AI industry just in time. AI has had a long winter. It repeated hype and disappointment for the last 60 years.

Nvidia entered the AI industry around the late 2000s when deep learning (one of the AI methods) leaped forward, driven by advances in hardware, algorithms, and data availability. That was just the right time. If Nvidia had tried to create an AI chip in the 60s, it apparently wouldn’t have worked. Even in the early 2000s, it was too early, so they would have just wasted their money.

Nvidia’s CEO Mr. Jensen got the indication of the AI revolution and invested in AI heavily at the right time. He said the gross profit rate ((sales-cost of sales)/sales) is not an indicator; it is just past information. When the investment succeeds, the gross profit rate is enhanced, but before that, you decide whether to invest or not. The readiness of AI doesn’t appear on financial statements when they invested in AI; it brought no sales immediately. After 10 years, suddenly ChatGPT showed up and changed everything. But before that, deep learning changed the landscape of the business world. He got feedback from the AI industry and kept increasing the investment.

By the way, Meta is not a 3D graphics creation company, which is why Meta’s metaverse graphics are so bad. It has no strength in that area, just a layman spending a crazy amount of money. It is like a bread company owner thinking “people in the future will get less bread because there will be too many people on Earth competing for food”, so the owner tries to invent hyper-abundant wheat. Really? You do that? Your family will get mad because you spend the hard-earned money.

On the other hand, Nvidia used what they already had (great know-how on GPU), extended it, and made an AI-focused GPU (still a GPU). They can use their resources and strength.