Talking Tech

Talking Tech is produced by the Progressive Tech Team of George O’Connor, Ian Robertson and Gareth Evans. Our aim is to bring you up to date with the tech news cycle each week. We comment via blog and podcast on the slings and arrows of the sector at a time of huge change.

<< Back to Talking Tech archive

February 27, 2024

Nvidia is still a semiconductor stock

Nvidia’s Q4 FY24 results attracted a lot of attention and a lot of column inches.  Looking back at the results, the analyst conference call and the press coverage, much is focused on AI rather than the company.  Nvidia gives exposure to the AI revolution, but it can’t stop being a semiconductor stock, with all the volatility that can entail.


The very high margins, 76.7% in Q4, have not gone unnoticed by analysts and commentators alike.   But if we are going to actually analyse this, we look beyond gross margins at the gross contribution margin.  Each extra $1 of (AI) revenue from Q3 to Q4 contributed 84c of gross profit, and from Q2 to Q3 it was 86c. There are always other factors at play, but Nvidia is clearly making hay while the sun shines, with the very high margin AI products driving revenues up.

nvidia revenue and margin progression

Looking forward to Q2 and the rest of FY25, management suggests that margins track back to the mid-70s%. But by semiconductor industry standards, and Nvidia’s, that is still far from normal. With gross margins declining, a lot of top line growth is required to keep the strong upwards earnings trajectory – particularly if operating costs have started to build an upward momentum.

But that’s clearly not a problem, as market consensus for FY25 and FY26 gross margins are 75.8% and 75.4%, respectively.

Position in the value chain

Nvidia depends on one major supplier more than any other: TSMC, the Taiwanese chip foundry, which actually makes and processes Nvidia’s core pieces of silicon.  Nvidia has a long-term relationship with TSMC. However, you can’t help but feel that some TSMC investors may feel a bit aggrieved, wondering how their world-leading company is not getting its dues – both in terms of money and kudos.



Nvidia’s top customers are Microsoft, Meta, Amazon and Google. They are also some of the few companies on the planet with the wherewithal and the guts to create something to challenge Nvidia, and they all have their own AI chips. However, this is more a question of developing something for themselves on the less demanding inference side of things, i.e. the application of what has been trained rather than the training process.

At present, the centre of attention is training. However, Nvidia estimates that about 40% of its FY24 datacentre revenue was for inference applications. The market was pleasantly surprised by this. While this suggests that Nvidia’s dominance in training is currently being replicated in inference, the reactions suggest that the market hasn’t got much insight into what exactly is going on.

As talk is increasingly about potential competitors and breaking the monopoly, attention naturally focuses on CUDA, suggesting that Nvidia’s software technology platform, which allows engineers to (in broad terms) program Nvidia’s chips, is a substantial barrier to entry.

AMD’s MI300X is said to be ahead of Nvidia’s current H100 flagship but there is little to suggest that AMD can get over the ease of use, familiarity and compatibility advantages of CUDA, or that AMD can get enough MI300Xs made or that MI300X will still have any edge over Nvidia’s soon to be released H200.  Looking at an industry that is, in effect, panic buying and where the mantra is that no one is going to get sacked for buying Nvidia, the race is Nvidia’s to lose.

Even without the threat in inference, there is the obvious risk that, as with many first wave spends (railways, canals, internet, big data et cetera), no one makes any money by buying Nvidia. But who will be the first to admit that their major corporate AI spend has yielded no financial returns?  Nobody this side of Christmas.

And even if that comes to pass, Nvidia would still be best placed to ride the second, profitable wave.

Is anyone listening to Jensen?

One of the oddities of the coverage is that, while everyone is getting excited about AI, Nvidia CEO Jensen Huang’s excitement isn’t just about AI.  AI is but a milestone for him.  He sees the whole world opening up, with accelerated processing (i.e. Nvidia and its GPUs) taking over from the hegemony of CPUs (traditional general central processors from Intel et al.), and with his software business permeating across the software universe too.

Never mind the application, look at the stock

Nvidia has long been a favourite play stock for tech investors, but there is an increasing interest from retail and generalists.  Semis stocks are not what they seem.

Nvidia’s revenue guidance is only given for one quarter and not for the year. It takes several months to make a semiconductor device and Nvidia (or rather TSMC) is booked out. We can therefore look forward to another excited set of Q1 results in May, with a beat that isn’t a beat. What matters then is guidance, and with the new H200 coming onstream in Q2, demand should not be an issue.

Away from the numbers, Jensen Huang is warming everyone up for his keynote speech at the GTC AI Conference in March. There is going to be plenty of newsflow and noise.

Investors should not, however, interpret the ability to hit / beat short-term figures as an ability to do anything with certainty in the medium term.

Things should be more predictable now that Nvidia has gone from being a consumer driven stock to a corporate spend driven one.  All may look set fair for a few months and Nvidia has a strong position in an exciting growth market, but a smooth ride is not guaranteed. Nvidia is still a semiconductor stock.

Ian Robertson

This communication is provided for information purposes only, and is not a solicitation or inducement to buy, sell, subscribe, or underwrite securities or units. Investors should seek advice from an Independent Financial Adviser or regulated stockbroker before making any investment decisions. Progressive Equity Research Ltd (“PERL”) does not make investment recommendations.

Opinions contained in this communication represent those of PERL and/or our affiliates at the time of publication and PERL does not undertake to provide updates to any opinions or views expressed. PERL does not hold any positions in the securities mentioned in this communication, however, PERL’s directors, officers, employees, contractors and affiliates may hold a position,  and/or may perform services or solicit business from, any of the companies or related securities mentioned.

Any prices quoted in our research are as at the previous day’s close.