Quick note today as we are dealing with the full brunt of flu and covid season right now at D2D HQ.

On the latest episode of The Circuit, we talked a lot about Intel’s diminishing significance in the data center market and corresponding rise of Nvidia’s influence. Last month, we ran the numbers and found that Nvidia was capturing over 70% share of data center processor spend. And it occurred to us that we might be able to take a guess at Nvidia’s revenue based on the limited data we have on the market from Intel’s and TSMC’s earnings.

Our first instinct was to compare Intel’s data center revenue with TSMC’s HPC segment, which loosely corresponds to data center products. Below we show the comparison.

The most interesting facet of this data is the way in which Intel has gone from doing double TSMC’s data center revenue to barely half in three years. Not surprising in hindsight, but the scale of the collapse still stands out.

Next we looked at the relationship between Nvidia and TSMC.

Lately, when asked about TSMC’s results, we have taken to saying that whenever their management speaks about AI, that should be read as code for Nvidia. A large portion of Nvidia’s cost of goods sold goes to TSMC. This is not an exact match as not all of Nvidia’s spend at TSMC goes for data center products, Nvidia still sells a lot of PC GPUs, but we think the math still works out roughly.

Last quarter, Nvidia’s COGS was 65% of TSMC’s HPC revenue, again not a perfect relationship, but a workable proxy. If that figure holds in Nvidia’s latest quarter, then it spent $5.4b on COGS. Assume Nvidia’s gross margins are the same this quarter as last, that then works out to $21 b in revenue for Nvidia. Current consensus estimates for Nvidia are looking for $20.2 billion, implying a decent size beat for the company. Taking this a step further, Nvidia’s share of TSMC’s HPC revenue and its gross margins have been on an upward trend for the past year, and there are no reasons to think that trend breaks this quarter. So if Nvidia’s share of TSMC goes to 70% and gross margins hit 78% that works out to $28 b in Nvidia revenue. That would be a staggering beat, from a company that has had several quarters of similarly massive revenue upside in recent quarters.

Of course, this is all very rough. We are taking shallow data patterns to make these predictions, and there are a lot of moving parts. Only so many companies can afford Nvidia’s products in bulk and if one of them delays shipments by a few months that would put a meaningful dent in these numbers. It is also incumbent on us to remind everyone that Nvidia has a bad track record when it comes to forecasting its quarters. Over the past ten years or so, they have had a half dozen or so meaningful misses. So take all of this with a grain of salt.

That being said, we have not heard a single thing that would lead us to expect Nvidia to miss this quarter. Eventually this will happen, but there is no reason to think the miss is coming any time soon.