Nvidia’s Latest Trick: A Dash of Genius, What!

by

Now, one gathers that the past year or so has been something of a rollercoaster for those chaps invested in Nvidia (NVDA 0.44%). A bit of a wobble, really, after a positively ripping good run. The stock took a bit of a dip – a rather alarming 37%, if one is to be precise – before, with a commendable show of spirit, bouncing back and reaching for the skies once more. All this fuss over inflation, tariffs, and the rather vague anxieties surrounding Artificial Intelligence – a bit of a kerfuffle, wouldn’t you say?

Still, Nvidia has done rather well for itself, hasn’t it? A gain of 977% over the last three years is nothing to sneeze at. Their graphics processing units – GPUs, to the knowing – have become the very gold standard for this AI business. And after such a breathless sprint, a little pause for refreshment is perfectly understandable. The stock currently sits a modest 12% below its recent peak, which, frankly, seems a perfectly reasonable place to be.

A new year, naturally, presents new opportunities. And it appears Nvidia is setting the stage for another little jaunt upwards. At a recent shindig – the Consumer Electronics Show, in Las Vegas, you know – their CEO, the rather dynamic Mr. Jensen Huang, made an announcement that might just cause a stir. A bit of a game-changer, perhaps, for those of us with a vested interest.

You Say You Want Chips?

The aforementioned CES is the place to be for all things technological. Gadgets, innovations, and this AI malarkey are all on display. Mr. Huang, a bit of a rock star in these circles, is a frequent keynote speaker. And this year was no exception. He does have a knack for the dramatic, that fellow.

And what did he announce? A most agreeable surprise, actually. Nvidia’s next-generation AI chip – the Vera Rubin, a rather dashing name, don’t you think? – is now in full production, a full six months ahead of schedule! Quite the feat, what!

“Vera Rubin is designed to address this fundamental challenge that we have: The amount of computation necessary for AI is skyrocketing. Today, I can tell you that Vera Rubin is in full production.”

Apparently, this Rubin architecture reduces the cost of processing AI ‘tokens’ by a staggering 90%, using a mere 75% of the GPUs previously required. A dashedly clever bit of engineering, wouldn’t you agree?

Nvidia already had a rather comfortable lead in this semiconductor game, releasing new processors every year – a positively brisk pace, compared to the usual two-year cycle. This relentless innovation – announced late in 2023 – has catapulted them ahead of the competition, making it rather difficult for the others to catch up. A bit of a pickle for them, really.

The revelation that the Rubin chip was already rolling off the production line took the industry by surprise, extending Nvidia’s already comfortable advantage. They had previously announced full production in the second half of 2026, so this is rather a swift bit of work.

One shouldn’t underestimate the implications, you see.

The Gathering Storm of Competition

The enthusiastic adoption of AI has led to a rather persistent shortage of chips with the necessary computational horsepower. Nvidia’s state-of-the-art chips have been in short supply, forcing customers to seek alternatives. A bit of a scramble, if you will.

And that brings us to the rather looming threat of competition. Several well-heeled technology companies are already developing rival processors. A bit of a free-for-all, really.

  • Advanced Micro Devices is making a splash with its MI350X data center GPU. They’ve just announced the MI440 series, which will be shipping shortly. A commendable effort, though whether it will be enough remains to be seen.
  • Broadcom leads the way in application-specific integrated circuits (ASICs), which can be customized for efficiency. They’re making inroads as a more energy-efficient alternative. A sensible approach, one might say.
  • Alphabet has been developing custom Tensor Processing Units (TPUs) for internal use for over a decade. Rumors abound that they may lease these chips to Meta Platforms, which would be a first. A rather bold move, if true.
  • Late last year, Amazon announced the latest version of its Trainium chip, claiming lower AI training costs than GPU alternatives. A perfectly reasonable proposition, of course.

There are others, naturally, but one gets the point.

Loading widget...

The Implications for the Discerning Investor

Despite the rather vigorous competition, Nvidia has maintained its lead by delivering best-in-class chips long before its rivals. The announcement that Vera Rubin chips are in full production only increases that advantage. A rather comfortable position to be in, wouldn’t you agree?

Nvidia also has a remarkably clear view of its future sales. Late last year, Mr. Huang stated that their backlog exceeded $500 billion, to be filled over the next six quarters. A rather substantial sum, wouldn’t you say?

Since then, Nvidia has reported revenue of $57 billion and forecast $65 billion for the next quarter. That suggests potential sales of as much as $378 billion next year – a growth of 155%. A rather optimistic projection, perhaps, but not entirely unreasonable.

Even that might be conservative. At a recent investor event, CFO Colette Kress said demand has increased since the $500 billion estimate, and they will “definitely” surpass it. A most encouraging sign, wouldn’t you say?

Nvidia’s industry-leading position, relentless innovation, and growing backlog suggest a long runway ahead. And at just 24 times next year’s expected sales, one might argue the stock is a steal. A perfectly sound investment, wouldn’t you say?

Read More

2026-01-17 11:13