NVIDIA: AI & cloud soaring, Turing launched – crypto-mining dead?
NVIDIA CEO Jensen Huang

NVIDIA: AI & cloud soaring, Turing launched – crypto-mining dead?

GPU and AI giant NVIDIA has just reported strong, analyst-beating Q2 results – and yet its share price fell. The takeaway for some was that the cryptocurrency mining market is dead – at least for NVIDIA. The reality is rather different, says Chris Middleton.

In-depth analysis for decision-makers

Internet of Business says

NVIDIA makes hardware products that are core to a number of booms: artificial intelligence and machine learning; driverless cars and autonomous vehicles; cloud data centres; virtual and augmented reality; gaming; and cryptocurrency mining.

Its high-end graphics processing units (GPUs) provide the massive number-crunching ability that powers these and other technologies, and it has long specialised in providing dedicated solutions in a business that grew out of helping PCs to display graphics better and faster.

The company has just announced what it called “the biggest leap in GPU architecture in over a decade” – the Turing ray-tracing GPU (see below) – and reported better than expected Q2 financials. And yet its stock fell amid reports that it had called an end to the cryptocurrency boom.

But had it?

First, those financials and what they reveal about a number of booming global markets.

Record results

NVIDIA reported total Q2 sales of $3.12 billion, up 40 percent year on year, with earnings per share up 91 percent compared with the same quarter last year, beating consensus Wall Street predictions.

The strong results were led by data centre and gaming, with visualisation and automotive hardware also hitting record levels, both sequentially and year on year, according to CFO, Colette Kress.

“Gross margins grew nearly 500 basis points year on year, while both GAAP and non-GAAP net income exceeded $1 billion for the third consecutive quarter. Profit nearly doubled,” she said.

Game on

Gaming revenue stood at $1.8 billion, up 52 percent year on year and five percent sequentially.

GPU revenue grew 40 percent from last year to $2.66 billion, while revenue from the company’s Tegra processor also grew 40 percent, to $467 million.

At $760 million, data centre revenues were significantly smaller than gaming, for example, but represented 83 percent year-on-year growth and an increase of eight percent sequentially.

In other words, NVIDIA is benefiting from the same boom in enterprise cloud services that has been powering recent results from Microsoft, Google, Amazon, and others.

And at the centre of that boom is artificial intelligence.

AI with everything

“This performance was driven by hyper-scale demand as internet services used daily by billions of people increasingly leverage AI,” confirmed Kress, adding that NVIDIA GPUs power real-time services, such as search, voice recognition, voice synthesis, translation, recommender engines, fraud detection, and retail applications.

“We also saw growing adoption of our AI and high-performance computing solutions by vertical industries, representing one of the fastest areas of growth in our business.”

Sectors ranging from oil and gas to financial services and transportation are harnessing the power of AI via NVIDIA’s “accelerated computing platform” to turn data into actionable insights, she said.

NVIDIA also powers five of the world’s seven fastest supercomputers, including the Summit AI at Oak Ridge National Laboratory, which currently holds the number one spot. Summit is powered by over 27,000 Volta Tensor Core GPUs.

Seeing the future

NVIDIA’s pro visualisation revenues grew to $281 million, up 20 percent year on year and 12 percent sequentially, driven by demand for real-time rendering and mobile workstations – as well as by AI and VR.

“These emerging applications now represent approximately 35 percent of pro visualisation sales,” said Kress. “Our strength extended across several key industries, including healthcare, oil and gas, and media and entertainment.”

Key wins for NVIDIA in the quarter include Raytheon, Lockheed, GE, Siemens, and Phillips Healthcare.

So the takeaway thus far from the results is that supporting the enterprise cloud in the data centre is the emerging hotspot in NVIDIA’s business, and AI is making its presence felt across every aspect of operations – including in the server closet.

“AI adoption continues to seep from one industry into another,” said CEO Jensen Huang. “The automation that’s going to be brought about by AI is going to bring productivity gains to industries like nobody has ever seen before.”

Motoring ahead

Finally (for the purposes of this analysis), NVIDIA’s automotive revenues hit a record $161 million, up 13 percent year on year and 11 percent sequentially.

“This reflects growth in our autonomous vehicle production and development engagements around the globe, as well as the ramp of next-generation AI-based, smart cockpit infotainment solutions,” explained Kress.

“There are two dynamics that are happening that are abundantly clear and will transform that industry,” said Huang. “The first, of course, is ride hailing and ride sharing.

“Those platforms – in order to make a recommendation of which taxi to bring to which customer – face a really large computing problem. It’s a machine learning problem; it’s an optimisation problem at very, very large scale. And in each and every one of those instances, you need high-performance computers to use machine learning to figure out how to make that perfect match, or the most optimal match.

“The second is self-driving cars. Every single car company that’s working on robot taxis or self-driving cars needs to collect data, label data, train on your network, and run those neural networks and cars. Every single one of them will need even more GPU-accelerated servers.

“And that’s just for developing the model,” he continued. “The next stage is to simulate the entire software, because we know that [motorists] travel 10 trillion miles per year. And the best we could possibly do is to drive several million ‘normal’ miles [to test driverless systems].

“What we really want to do is to be able to simulate and stress-test our software stack. And the only way to do that is in virtual reality. So that’s another supercomputer that you have to build for simulating all your software across those billions and billions of virtually created, challenging miles.”

In July, Daimler selected NVIDIA’s Drive Pegasus system as the AI brain for its Level-4 and -5 autonomous fleets. The collaboration brings together NVIDIA’s AI and self-driving platforms, Bosch’s hardware and systems, and Daimler’s vehicle expertise and global brand.

More than 370 companies and research institutions are now using NVIDIA’s automotive platform, claimed Kress. “With this growing momentum and accelerating revenue growth, we remain excited about the intermediate and long-term opportunities for autonomous driving business,” she said.

The new Turing test

Next: that great leap forward.

NVIDA formally announced its Turing architecture at SIGGRAPH last week, where it also unveiled the first Turing-based processors – the Quadro RTX 8000, 6000, and 5000 GPUs – and the RTX Server.

These introduced AI-enabled ray-tracing – the ability to more accurately model and render how light behaves in the real world, bringing new levels of photorealism to pro visualisation, CGI, VR, and other visually based technologies.

CEO Huang said, “Turing is a reinvention of computer graphics; it is the first ray-tracing GPU in the world; it’s the first GPU that will be able to ray-trace light in an environment and create photorealistic shadows and reflections, and be able to model things like real lights, global illumination, and indirect lighting.”

“There are two major ways that we’ll experience the benefits [of Turing] right-away,” he continued.

“The first is, for the markets of visualisation today, they require photorealistic images. Whether it’s an IKEA catalogue or a movie, or architectural engineering, or product design, or car design… all of these markets require photorealistic images. And the only way to really achieve that is to use ray-tracing with physically based materials and lighting.

“The technology is complicated, and has been computing-intensive for a very long time. It wasn’t until now that we’ve been able to achieve it in a productive way.

“That marketplace is quite large and it’s never been served by GPUs until now. All of it has been run on CPU render farms – gigantic render farms in all these movie studios and service centres.”

Then Huang said enigmatically, “The second area where you’re going to see the benefits of ray-tracing, we haven’t announced yet.”

Given the spread of NVIDIA’s business, we could speculate that it is likely to be in the area of virtual reality or immersive gaming – but there are no details as yet.

Which brings us to the cryptocurrency enigma, which Huang has – finally – been rather more open about.

The big crypto problem

In recent years, NVIDIA’s fortunes have been more closely entwined with cryptocurrency mining than the company has liked to admit.

As explored in a previous Internet of Business report, it is no coincidence that NVIDIA’s share price has more or less exactly tracked the growth curve of cryptocurrencies over the past four years.

Amateurs and professionals alike have hoarded GPUs to create mining rigs, pushing up the retail price of many – which has risked making data centres, and therefore cloud services, more expensive to run – while reducing core market product availability.

Over the past few years, despite developing some dedicated solutions, NVIDIA has barely acknowledged crypto-mining in some financial results, talked down its impact in others, walled it off within discrete reporting, or suggested that it is something that gamers do in their spare time (which is true).

In other words, the message has been: if miners use NVIDIA hardware designed for another purpose, then so be it.

The reason for Huang’s caution is completely understandable: the crypto market is diverse, volatile, unstable, unpredictable, controversial, and constantly evolving, and so no sensible business leader would want to tie a company’s fortunes too closely to it – while still allowing the busness to reap the benefit of increased sales and prices, of course.

Cost per watt

On the buy side of the equation, miners have long faced a tough problem: the cost per watt of mining for digital tokens is critical to whether they can make real money from cryptocurrencies – once they have invested in rigs that may cost thousands of dollars and consume large amounts of electricity.

Our report on the twin-GPU-powered Qarnot heater/currency-mining rig explored this problem in depth, and found that few – if any – buyers would make money from the device at ether’s then value. But at least they would have a source of heat.

So what has NVIDIA got to say about cryptocurrency mining today?

“Our revenue outlook had anticipated cryptocurrency-specific products declining to approximately $100 million, while actual crypto-specific product revenue was $18 million,” said Kress.

“We now expect a negligible contribution going forward.”

Huang added to his CFO’s unexpected revelation by saying, “It is the case that we benefited in the last several quarters from an unusual lift from crypto,” finally acknowledging NVIDIA’s historic windfall from currency mining.

“At the beginning of the year, we thought – we projected – that crypto would be a larger contribution through the rest of year. But, at this time, we consider it to be immaterial for the second half.”

Pressed on this point by analysts, he added, “It’s hard to estimate, no matter what. […] The answer is we’re expecting – we’re projecting zero, basically.

“How much of GeForce [an NVIDIA GPU] could’ve been used for crypto? A lot of gamers at night, they could – while they’re sleeping, they could do some mining. So, did they buy it for mining or did they buy it for gaming? It’s kind of hard to say.

“Some miners were unable to buy our OEM products, and so they jumped onto the market to buy it from retail, and that probably happened a great deal. And that all happened in the last – the previous several quarters, probably starting from late Q3, Q4, Q1, and very little last quarter. But we’re projecting no crypto-mining going forward.”

The death of crypto?

Huang’s bleak comments have been interpreted by some investment analysts as forecasting the death of crypto-mining or digital currencies, for NVIDIA at the very least. In reality, Huang is – sensibly – distancing his company from the mining market yet again.

In an investment world that works on the principle of ‘the greater fool’, only a massive fool would believe that crypto is dead. And Huang is no fool at all.

The cryptomining and currency markets may have cooled, and some amateur miners may have realised that they’ve spent thousands of dollars on depreciating hardware while their electricity bills have soared (some may even have been tempted to hijack someone else’s processors), but it’s certainly not dead. Just look at the soaring levels of crypto-linked cybercrime:-

Indeed, some analysts believe that the next bubble or spike in cryptocurrency valuations may suggest something much more troubling than hype or hysteria: that the real-world economy is about to tank.

Are they right? Who can say with any certainty in the current world, in which bizarre and aberrant behaviour is fast becoming normalised.

Either way, no CEO would want to flag that up in a set of excellent quarterly results.