Nvidia’s stock has been on a roller-coaster ride since the company last reported earnings, plunging as low as $108 in October before rallying strongly in recent weeks.
Bulls are hoping the GPU giant’s October quarter (fiscal third quarter) report will drive additional gains. Among analysts polled by FactSet, the consensus is for Nvidia to report revenue of $5.78 billion (down 19% annually) and non-GAAP EPS of $0.71.
Nvidia typically provides quarterly sales guidance in its earnings reports. For the January quarter, the company’s revenue consensus stands at $6.07 billion (down 21%).
Eric Jhonsa, Real Money’s tech columnist, will be live-blogging Nvidia’s report, which is expected at 4:20 P.M. Eastern Time, along with an earnings call scheduled for 5 P.M..
Please refresh your browser for updates.
6:12 PM ET: Nvidia’s call has ended. Shares are up 1% after-hours to $160.76 after Nvidia reported mixed FQ3 results (revenue beat by $150M, but EPS missed by $0.13 due to a $702M inventory charge) and guided for FQ4 revenue of $6B (+/- 2%), a little below a $6.07B consensus at the midpoint.
Nvidia’s Data Center, Gaming and Automotive segment sales topped consensus estimates, while ProViz and “OEM & Other” sales fell short. On the earnings call, CFO Colette Kress said Nvidia expects Data Center, Gaming and Auto sales are each expected to be up slightly Q/Q in FQ4.
Data Center sales (up 31% Y/Y in FQ3) are expected to benefit in FQ4 from strong demand for Nvidia’s new H100 flagship server GPU, partly offset by soft Chinese demand. Gaming sales (down 51% Y/Y in FQ3) are expected to once more be hurt by inventory corrections, albeit with Nvidia expecting gaming GPU channel inventories to return to normal levels by quarter’s end.
Thanks for joining us.
6:03 PM ET: Huang makes some closing remarks. Notes Nvidia plans to capex roughly flat over the next few quarters. He talks up new product launches, Nvidia’s Data Center momentum and long-term adoption for accelerated computing.
6:01 PM ET: A question about whether networking supply is caught up with demand. And one about high stock comp (~$700M in FQ3).
Kress says supply has been getting better each quarter. Indicates networking supply is getting much better in FQ3/FQ4, while noting supply constraints for complementary products remain an issue.
Regarding stock comp, she says it’ll remain an important part of Nvidia’s employee compensation, while noting Nvidia’s reduced hiring activity will have an impact going forward.
5:57 PM ET: A question about the impact of crypto-related GPU demand.
Huang says Nvidia doesn’t expect blockchain to be a big part of Nvidia’s business going forward. Notes crypto miners can unload used Nvidia GPUs in the aftermarket. Declares that Nvidia’s recent Ada gaming GPU launches were “a home run.”
5:55 PM ET: A question about Nvidia’s plans to adjust how it labels hyperscaler sales going forward.
Kress says Nvidia’s Data Center customers are “complex.” That while there are only 7 or 8 hyperscalers, there are now other big cloud clients, and that enterprise adoption of cloud services featuring Nvidia GPUs is also growing fast.
Huang notes hyperscalers are both “sell-to” customers buying GPUs for internal workloads and “sell-with” customers providing cloud services to third parties. Adds that public cloud services are poised to become “the vast majority” of hyperscalers’ GPU consumption.
5:50 PM ET: A question about the expected pace of the H100 ramp.
Huang says “customers are clamoring” to ramp the H100 as quickly as possible due to its performance gains/TCO improvements. Adds that H100-based cloud services are expected to arrive in Q1.
5:47 PM ET: A question about Data Center’s compute (GPU-related) sales in FQ3. And one about Nvidia’s A800 GPU, which it’s selling to Chinese clients due to export restrictions on the A100/H100.
Kress says compute sales were “flattish” Q/Q in FQ3. Huang stresses the A800 is hardware-limited to make sure it complies with new U.S. export controls.
5:45 PM ET: Another networking question. Specifically, about how Nvidia’s InfiniBand offerings and Spectrum Ethernet switches are positioned.
Huang asserts the value of improved networking efficiency can’t be overstated from a cost standpoint, and that Nvidia is differentiating its networking offerings via software in addition to hardware.
5:42 PM ET: A question about expected networking growth.
Huang notes Nvidia’s networking sales are closely linked to the growth of high-performance computing workloads, and that it’s not looking to compete in the low end of the market. Mentions Nvidia’s 400G InfiniBand offerings power the AI-related Azure solutions it’s developed with Microsoft.
5:39 PM ET: A question about Nvidia’s Grace Arm server CPU (due to launch in 1H23).
Huang asserts Grace was “designed for very large data processing at very high speeds,” such as what’s needed for major AI workloads, and that it was engineered to work well with Nvidia’s server GPUs. Reiterates production samples will ship in Q1 2023.
5:36 PM ET: A question about Nvidia’s FQ4 sales guidance for various segments.
Kress says Nvidia expects Gaming, Data Center and Automotive sales to each grow Q/Q “just a little bit.” Adds that ProViz will remain a headwind.
5:34 PM ET: A question about whether Data Center sales are benefiting from Chinese customers stockpiling the products that they have access to.
Kress says Data Center sales tend to be driven by specific projects, and suggests Nvidia doesn’t see any indication of major stockpiling.
5:32 PM ET: A question about the FQ3 inventory charge.
Kress says the charge is almost entirely due to China, and how new export controls have changed its demand forecasts for Chinese sales. Adds that Nvidia’s Q/Q inventory growth has a lot to do with new product ramps.
5:30 PM ET: A question about how software revenue will ramp.
Huang says Nvidia is “making excellent progress” with its AI Enterprise offering. Notes today’s Microsoft announcement, and says it’s reflective of AI-related demand for Nvidia’s server GPUs among startups relying on public clouds. He also notes Omniverse and other offerings have been made available on public clouds, as part of an effort to ease adoption/deployment.
5:26 PM ET: A question about what Gaming sell-through now looks like. And one about cloud demand if cloud capex falls.
Kress says FQ3 Gaming sell-through was “relatively solid,” with desktop GPUs a little softer than notebook GPUs. Says Nvidia expects end-demand to pick up in FQ4 due to the holiday season and Ada GPU launches.
Regarding cloud demand, Jensen Huang once more talks up the secular trends of accelerated computing and AI workload adoption. Also highlights the impact of larger natural-language AI models and “generative AI” workloads that involve using AI to create content.
5:21 PM ET: The Q&A session is starting.
5:20 PM ET: With Nvidia’s FQ3 Gaming sales having come in at $1.57B, Kress’ comments suggest FQ4 Gaming revenue will be above a $1.48B pre-earnings consensus.
5:19 PM ET: Regarding FQ4, Kress says Gaming is expected to return to Q/Q growth. Data Center is expected to benefit from the ramp for Nvidia’s H100 flagship server GPU, and to be hurt by China softness.
5:18 PM ET: Kress says the $702M inventory charge Nvidia recorded in FQ3 was largely related to the new Chinese export controls.
5:16 PM ET: Automotive revenue is said to benefit from production ramps for design wins involving Nvidia’s Drive Orin SoC. Kress reiterates Automotive is set to become Nvidia’s next multi-billion dollar business.
5:14 PM ET: Regarding ProViz, Kress suggests inventory corrections will continue to weigh on sales in FQ4.
5:13 PM ET: Kress asserts once more that Nvidia remains confident about long-term gaming growth. Notes the popularity of major game launches/events, and that Steam recently hit a new record for concurrent users.
5:11 PM ET: Regarding Gaming (hit by major inventory corrections in recent quarters), Kress says Nvidia believes channel inventories are on track to return to normal levels by the end of FQ4. Adds that macro and China lockdowns remain near-term demand headwinds.
5:08 PM ET: Kress says networking product sales (included in Data Center) also saw strong growth, thanks in part to hyperscaler demand.
5:07 PM ET: Kress highlights Nvidia’s morning announcement about teaming with Microsoft to build a giant AI supercomputer on Azure, and that the solution will leverage Nvidia’s entire AI software stack.
5:05 PM ET: Kress is talking. She starts by recapping Nvidia’s FQ3 performance. Notes Data Center growth was driven by leading U.S. cloud providers and a broadening set of consumer Internet companies.
She adds that Data Center sales to “vertical industries” (traditional enterprises) were also good, and that Chinese demand broadly remains soft.
5:02 PM ET: IR chief Simona Jankowski is going over Nvidia’s safe-harbor statement. Typically, Nvidia’s calls have featured prepared remarks from CFO Colette Kress, after which Kress and CEO Jensen Huang both field questions from analysts.
5:01 PM ET: The call is starting.
4:58 PM ET: Nvidia ended FQ3 with $13.1B in cash/equivalents and $10.9B. The cash balance fell by $3.89B Q/Q, thanks in large part to buybacks.
4:56 PM ET: Nvidia’s earnings call should kick off in a few minutes. I’ll be covering it.
4:54 PM ET: Notably, Nvidia says the Chinese server GPU sales that it lost in FQ3 due to new export controls were “largely offset by sales of alternative products into China.” It’ll be interesting to hear what’s said on the call about expected demand for these products going forward.
4:51 PM ET: Nvidia on the potential impact of Ethereum’s shift to proof-of-stake on gaming GPU demand:
“We believe the recent transition in verifying Ethereum cryptocurrency transactions from proof-of-work to proof-of-stake has reduced the utility of GPUs for cryptocurrency mining. This may have contributed to increased aftermarket sales of our GPUs in certain markets, potentially impacting demand for some of our products, particularly in the low-end.”
4:48 PM ET: Like some other tech companies, Nvidia is slowing down its spending. The company’s non-GAAP operating expenses totaled $1.79B in FQ3 — up 30% Y/Y, but just 3% Q/Q and a little below guidance of $1.82B.
For FQ4, Nvidia is guiding for non-GAAP opex to fall slightly Q/Q to $1.78B. GAAP opex, which among other things includes stock comp, is guided to be at $2.56B vs. $2.58B in FQ3.
4:44 PM ET: In spite of recent write-downs, Nvidia ended FQ3 with $4.45B worth of inventory, up from $3.89B at the end of FQ2 and $2.23B a year earlier. Likely at play: Large purchase commitments made to TSMC and pending production ramps for new gaming and server GPUs.
Prepaid supply agreements (TSMC looms large here) were flat Q/Q at $3.14B. Nvidia also says it had $2.75B worth of “non-inventory purchase obligations,” of which $1.59B involved multi-year deals with cloud service providers.
4:40 PM ET: Due to OEM inventory corrections, ProViz revenue (driven by workstation GPU sales) fell 60% Q/Q and 65% Y/Y to $200M, below a $363M consensus.
Automotive revenue (benefiting from production ramps for Nvidia’s Drive ADAS/autonomous driving platform) rose 14% Q/Q and 86% Y/Y to $251M, topping a $229M consensus.
OEM & Other revenue (it includes sales of Jetson edge computing offerings and non-gaming GPUs, and is also hurt by inventory corrections) fell 48% Q/Q and 69% Y/Y to $73M, below a $128M consensus.
4:36 PM ET: Nvidia is now up 1.3% AH to $161.20. Rivals AMD and Intel are up 0.6%.
4:34 PM ET: The reason EPS missed estimates in spite of a revenue beat and buybacks: Nvidia’s non-GAAP gross margin was 56.1% — up from 45.9% in FQ2, but down from 67% a year ago and below guidance of 65% (plus or minus 50 bps).
In its CFO commentary, Nvidia notes its GM was hurt by a $702M inventory charge (this follows $1.32B worth of charges in FQ2). The company is guiding for its GM to rebound to 66% (plus or minus 50 bps) in FQ4.
4:27 PM ET: Notably, Nvidia spent $3.65B on stock buybacks in FQ3. That follows $3.34B worth of buybacks in FQ2.
4:26 PM ET: Gaming revenue fell 23% Q/Q and 51% Y/Y to $1.57B. But it topped a $1.42B consensus.
4:24 PM ET: Likely going over well with investors: In spite of the new China restrictions, Nvidia’s Data Center revenue came in at $3.83B — up 1% Q/Q and 31% Y/Y, and above a $3.72B consensus.
4:22 PM ET: Nvidia guides for FQ4 revenue of $6B, plus or minus 2%. That’s below a $6.07B consensus.
Shares are up 3.2% after-hours.
4:21 PM ET: Results are out. FQ3 revenue of $5.93B beats a $5.78B consensus. Non-GAAP EPS of $0.58 misses a $0.71 consensus.
4:10 PM ET: FQ3 consensus estimates for Nvidia’s product segments:
Data Center – $3.72B (+27% Y/Y)
Gaming – $1.42B (-56%)
Professional Visualization – $363M (-37%)
Automotive – $229M (+70%)
OEM & IP – $128M (-45%)
With near-term Gaming and ProViz expectations very low (both segments are also expected to see big Y/Y sales declines in FQ4), Data Center’s performance (hurt by China, but benefiting from strong U.S. cloud capex) is very much in the spotlight right now.
4:03 PM ET: Shares are going into earnings down 46% YTD. However, in spite of today’s decline, they’re up 47% from an October low of $108.13.
4:01 PM ET: Amid a broader tech-sector and chip-stock rout, Nvidia closed down 4.5% today to $159.10. The FQ3 report is expected to arrive at 4:20 ET.
3:57 PM ET: Of note: Nvidia previously warned that new export restrictions on server GPU sales to China could impact its FQ3 sales by up to $400M. Between this and ongoing gaming GPU pressures, informal expectations for Nvidia’s results and guidance might be below consensus estimates.
3:54 PM ET: The FactSet consensus is for Nvidia to report FQ3 revenue of $5.78B and non-GAAP EPS of $0.71.
For FQ4 (quarterly sales guidance should be provided in the report), the revenue consensus is at $6.07B.
3:51 PM ET: Hi, this is Eric Jhonsa. I’ll be live-blogging Nvidia’s earnings report and call.
Source: https://www.thestreet.com/technology/nvidia-third-quarter-earnings-live-blog?puc=yahoo&cm_ven=YAHOO&yptr=yahoo