At a time when Nvidia's (NVDA) stock is pricing in more good news than it was during the summer, the GPU giant's management is still striking an upbeat tone about both near-term demand trends and long-term growth opportunities.
Over the last 10 days, Nvidia CFO Colette Kress has made appearances at Barclays and Credit Suisse conferences, while Rob Csongor, the GM of Nvidia's Automotive business, has appeared at a Deutsche Bank conference. Here are some notable takeaways from the remarks that Kress and Csongor made at these events.
1. Demand From Cloud Giants Continues to Improve
About a month ago, Nvidia reported that its Data Center segment saw revenue drop 9% annually in its October quarter to $726 million. However, the company also reported that its sales to internet/cloud giants (the proverbial hyperscalers) rose sequentially and annually following an early-2019 slowdown, and forecast the Data Center segment would see "strong sequential growth" in the January quarter as hyperscaler demand continued rebounding.
Kress gave no indication that this trend had changed since mid-November: At the Credit Suisse conference, she said that Nvidia will "probably" see a higher sequential growth rate for the Data Center segment than the 11% growth seen in the October quarter.
Currently, the consensus is for Nvidia to post January quarter Data Center revenue of $829 million, up 14% sequentially and 22% annually.
2. Nvidia Suggests It Isn't Seeing a Big Impact from Intel PC CPU Shortages
When asked at the Credit Suisse conference about the impact of ongoing Intel (INTC) PC CPU shortages on Nvidia's gaming GPU business, Kress indicated that Nvidia didn't see a major impact during its October quarter, since Nvidia's gaming GPU sales skew towards high-end PCs and the Intel shortages had generally involved cheaper CPUs. She added -- amid recent signs that some high-end CPUs are also now supply-constrained -- that Nvidia has already "taken into account what we know" about the shortages and incorporated it into its January quarter guidance.
Kress also downplayed concerns about the shortages at the Barclays conference. "We believe we're in a good position, being positioned in the high end, to have those overall CPUs," she said. "But we'll see how this overall quarter comes through."
Kress reiterated that Nvidia expects its notebook gaming GPU revenue to drop sequentially during the January quarter (the company has attributed this to lower seasonal builds), and that it expects desktop gaming GPU sales to rise.
3. Nvidia Believes the Auto Industry's Woes Have Contributed to Autonomous Driving Delays
"The automotive industry though is having some challenges in terms of their overall financial situation," Kress said at the Barclays conference, while adding that it will probably be "a little bit more than a year" before Nvidia sees volume shipments for autonomous driving solutions that go into production cars.
Kress also suggested that a major design win with Volvo for a Level 2+ autonomous driving solution now won't see volumes ramp until 2022 -- CEO Jensen Huang had previously talked about shipments starting in late 2020 or early 2021 -- and that other automotive wins will also start ramping around 2022.
Csongor offered a similar view of things. Though noting Level 5 autonomous cars requiring no human intervention are arriving sooner since they're only meant to be driven in specific areas (think Waymo's current Phoenix test), he argued that "Germany and Dieselgate and some of the [auto industry's] financial conditions" have led to delays for Level 2+ cars that still require human drivers in many situations, but which can be used anywhere.
"So, I think roughly 2022, 2023 is when you'll see an inflection of designs that where you get some ramps for those Level 2+ designs," Csongor said. He added that Level 2+ systems will essentially be capable from a hardware standpoint of supporting more advanced Level 4 autonomy, and that Tesla (TSLA) , whose latest Autopilot hardware relies on proprietary silicon, is motivating other automakers to develop rival autonomous solutions with Nvidia's help.
4. Software Revenue Streams Are a Possibility for Nvidia's Automotive Business
When asked if Nvidia, which has made large software investments for its Drive autonomous driving platform, could see "non-silicon revenue streams" for its Automotive business over time, Kress didn't rule out the possibility.
"[Long] term, when you think about other types of revenue streams, we absolutely have that possibility from a software perspective to think about different ways to overall monetize and continue to add value throughout the overall product life," Kress said. She also noted that Nvidia's GPUs will help power the data center infrastructures needed by automakers to do things such as simulate autonomous driving activity.
Csongor, for his part, noted that providing continued software updates to cars featuring Nvidia's silicon is a part of the company's automotive strategy. "It's inconceivable that a software-defined computer would not have updates, right?," he said. "[One] of the things that makes Nvidia special is the fact that the platform is software-defined."
5. Nvidia Says to 'Stay Tuned' Regarding Next-Gen GPUs
At both the Credit Suisse and Barclays conferences, Kress was asked about Nvidia's much-rumored plans to launch a next-gen GPU architecture -- it's believed to be codenamed Ampere and rely on one or more 7-nanometer (7nm) manufacturing processes -- next year. And in each case, she was quite careful about what she did and didn't say about Nvidia's GPU roadmap.
"We are always busy at work building our overall architectures, whether that be our architectures as a whole and it tends to serve many of our different markets, whether that be gaming or [professional visualization] or overall data center. So fear not, we're working on our technology," Kress said at the Credit Suisse conference. "Our innovations and our process technology has been exceptional over these last years. So stay tuned."
Kress also echoed Huang's past remarks about new server GPU platforms requiring close collaboration with hyperscalers and other major customers to roll out, and that its large software investments shouldn't be overlooked when gauging the competitiveness of its GPUs.