There has been a lot of noise on our favorite company Nvidia (NVDA), including an Earnings Watch Party at a NYC bar, which was hilarious and got me thinking of past markets. In the Roaring Twenties (1920-1929), before the infamous market crash of 1929 – JP Morgan said towards the end that if your shoeshine boy is giving you tips you know the market is completely overbought. In Peter Lynch’s (the highly succesful Fidelity Magellan fund manager),time (1977-1990), it was the rude guests at cocktail parties who would give him tips before Black Monday in Oct 1987. And in my own experience during Harshad Mehta’s time (1989-1992) India’s first major bull market, the cab driver recommending SBI (State Bank of India).
I don’t believe we’re in a bubble, but the index is (including the M-7) is and has been overpriced for a while – we’ve discussed that several times. We will have more days like the Japanese carry trade crash, that’s just how volatility derivative trading works. Similarly we’ll also get violent upswings.
We pretty much discussed in the group that Nvidia was prized to perfection and clearly it would have been tremendously optimistic on anyone’s part to expect a bump, post earnings – there were even a few good jokes about that. In fact, one of you was sharp enough to short prior to earnings and made a decent return.
But regardless of all this euphoria/noise and quarterly gyrations, let’s take a brief look at Nvidia‘s fundamental’s related to Blackwell.
Blackwell, their latest iteration is the most expensive, and complex GPU yet, and it’s in extremely high demand for the next 18 months.
Not surprisingly, quite a few news outlets and financial platforms highlighted the delay in Q4 shipments of Blackwell to Q1 2025, citing design flaws.
For an extremely difficult product that is shipping to the market in record time, this was expected and the supposed design flaw is related to packaging, which too was buried under the headlines. ASML’s EUV, which is the backbone of advanced AI chip making technology has gone through these quarterly delays often.
Nvidia emphasized that there will be several Billion dollars of Blackwell revenue in Q4, before it ramps production in the next year. Importantly, their current GPU Hopper, is easily filling demand for Blackwell, and then some.
The next platform has a very nice article with Nvidia’s CFO, if you get a chance to read it – it doesn’t have a paywall.
Bottom line – Ignore the volatility, Nvidia is still going to do extremely well in the next 5 years, even with competitors inching their way in and questions regarding monetization of AI – the cloud service providers alone have committed $150Bn next year in spending on AI.
It will remain overpriced by most yardsticks and benchmarks, and obviously for us investors the next 5 years of returns will be comparatively muted, but that hardly erodes my confidence in the company. I plan to buy on dips – I last bought at $104, and if it falls to that level or lower, I’m likely to add and will alert if I do so. In the next few months I will also dispose off 5-10% should it go to $145-150.