Nvidia grew Q3 revenue and profits as inventory clouds gather

Nvidia reported its financial results for the third quarter of its fiscal 2019 today. The company brought in revenue of $3.18 billion, up 21% from this time last year, and operating income of $1.06 billion, up 18% from a year ago. The company also posted net income of $1.23 billion, up 47% from a year ago, and diluted earnings per share of $1.97, up 48% from this time in 2017. Gross margin rang in at 60.4%, up 0.9 percentage points from this time last year. The company's stock fell about 16% in after-hours trading on those numbers, thanks to a miss on Wall Street analyst estimates.

Q3 FY19 Q3 FY18 Change
Revenue $3.18 billion $2.64 billion Up 21%
Gross margin 60.4% 59.5% Up 0.9 pp
Operating income $1.06 billion $895 million Up 18%
Net income $1.23 billion $838 million Up 47%
Diluted earnings per share $1.97 $1.33 Up 48%

Nvidia unsurprisingly continues to make the lion's share of its revenue from its GPUs. Across all of its GPU-related businesses, the company brought in $2.77 billion in revenue, up 25% from a year ago. GeForce-related revenue was $1.76 billion, up 13% from a year ago, although the company acknowledges that revenue for its gaming business in the wake of its Turing launch was "short of our expectations."

The gaming business' results were counterweighted by softening seasonal demand for Tegra SoCs for the Nintendo Switch. Tegra products in general brought in $407 million in revenue, down 3% on the year, across automotive, Nintendo, and embedded edge-computing products.

The company's professional visualization business brought in $305 million in revenue, up 28% from this time a year ago and a result Nvidia describes as a record. The company touted strong demand for its desktop and mobile workstation products in this segment. Data-center revenue was $792 million, up 58% from a year ago, on continued strong sales of Volta V100 and DGX systems. Nvidia also says its new Tesla T4 accelerator gave its data-center results a boost.

Nvidia's automotive business was $172 million, up 19% from a year ago, accounting for sales of infotainment hardware, Drive PX systems, and software development agreements with partner companies. OEM and IP revenue was $143 million, down 23% on the year, thanks to the evaporation of demand for GPUs for cryptocurrency mining.

For the fourth quarter of Nvidia notes that "excess channel inventory of mid-range Pascal products" will weigh on its fourth-quarter results, although it expects that excess inventory to clear out of retail and e-tail in "one to two quarters." Nvidia says it intends to ship no "meaningful" numbers of mid-range Pascal cards into the channel for the fourth quarter to allow demand to draw down those stocks. With all that in mind for the fourth quarter of its fiscal 2019, Nvidia expects revenue of $2.7 billion, plus or minus 2%, as well as GAAP gross margin of 62.3%, plus or minus half a percentage point.

Comments closed
    • tipoo
    • 1 year ago

    On a tangent, I didn’t realize that campus render from the home page picture was actually built and working by now, not just a concept render

    [url<]https://www.youtube.com/watch?v=iCKx8LidOdM[/url<] [url<]https://www.youtube.com/watch?v=fybjbV0lndY[/url<]

    • Anonymous Coward
    • 1 year ago

    Interesting that nVidia talks about excess stock, AMD just launched the RX590, suggesting to me that they don’t feel too pressured by old inventory. I mean, RX590 hardly qualifies as a new product, yet it clearly isn’t just a new sticker on the old product.

    • Krogoth
    • 1 year ago

    Nvidia is about to unleash its 1060Ti (basically artificially binned GP104s so they can move inventory) in response to RX 590 and do aggressive price wars on existing GP106 and GP104 SKUs for this upcoming holiday season.

    They’ll try to clear the excessive post-cryptocurrency bubble inventory to make room for upcoming lesser Turing SKUs.

    Nvidia projected too far on Turing launch. The only buyers would have been a small minority of hardcore users who get the latest, greatest no matter what. The rest of discrete market isn’t too enticed with current Turing SKUs at their current price points. They already are running on Maxwell through Pascal-era hardware so they aren’t exactly in a massive rush to upgrade either. DX12-era content is still a small minority which is only time where Turing shines over Pascal/Maxwell.

    • NovusBogus
    • 1 year ago

    I am shocked–truly shocked, I say–that a quarter marked by the launch of a product line that struggles to outperform last-gen inventory that costs less would yield unimpressive revenue.

      • Beahmont
      • 1 year ago

      Unimpressive revenue? In what world?

      Their revenue, margins, and profit were all up compared to last year still. They just sold X-1 million units instead of X million units like they though they would.

        • jihadjoe
        • 1 year ago

        In the financing world it’s unimpressive, which caused Nvidia stock to drop 15%.

        [url<]https://www.thestreet.com/investing/stocks/nvidia-tumbles-on-soft-guidance-5-key-takeaways-14784022[/url<]

          • DavidC1
          • 1 year ago

          The financing world basically expects every sector they are into grow indefinitely and at eye popping rates.

          At some point it’ll slow down, and even go the other way.

            • NoOne ButMe
            • 1 year ago

            Except the Wall Street numbers, at least the ones in the link in the article, look suspiciously like Nvidia’s own estimates in their 2Q2019 call.

            Nvidia missed both in revenue, and in margins to their own projections. If Nvidia had just hit it’s margins, it would have had the EPS that the market predicted. Although probably still lose some stock price due to missing revenue.

          • K-L-Waster
          • 1 year ago

          It’s not dropping on the results, it’s dropping on the forward guidance.

        • derFunkenstein
        • 1 year ago

        They were juuuuuuuuuust under their targets, so of course that’s what warrants a temporary drop in stock price.

        Their targets were outlined in the last quarter’s news post here:

        [url<]https://techreport.com/news/34001/nvidia-marks-the-death-of-crypto-demand-in-q2-of-its-fiscal-2019[/url<] [quote<]For its next quarter, Nvidia expects $3.25 billion in revenue, plus or minus two percent. That figure includes no income from crypto demand. Going by Nvidia's third-quarter fiscal 2017 results, that figure would represent a 23.3% year-on-year increase, suggesting the company's meteoric rise of late might be tapering off a bit. GAAP gross margin is projected at 62.6%, which would reflect a 3.1-percentage-point increase. With the release of new GeForce products imminent, we'll have to see just how much pent-up demand for next-generation gaming products the company is able to unleash as cryptocurrency demand finally seems to be dying off.[/quote<] Like they missed hitting the "minus two percent" part of that by $5M, and missed the middle of that range by like $70M. That's pocket change compared to the 3.2-ish billion they brought in, but all the same they missed their target. Wall Street is ruthless.

    • rudimentary_lathe
    • 1 year ago

    Down 17% after hours on a significantly weaker than expected revenue forecast. I can’t say I’m upset that the stock is taking a hit. They kinda deserve it after the huge price hikes with the Turing cards.

    You’d think with excess inventory of Pascal cards we’d see more price discounting going forward. Or maybe they’ll just hold off on the mid-range Turing cards until the Pascal cards get cleared out. We should find out soon enough with Black Friday fast approaching.

    That said I see little reason to go with the Green team in the mid-range since they still don’t support FreeSync. AMD still has a competitive and reasonably priced card in the RX 480 8GB.

      • Johnny Rotten
      • 1 year ago

      I mean its been really rehashed to death but I think nvidia really misjudged the market. Fact of the matter is no one with 1080p monitors with common sense would go near the 20×0 line. Complete and total overkill, especially given the price. 1080p is the realm of low-mid cards. 1440p is well served by 1070-1080/rx590/vega56 line. Really you’re looking at 3440×1440 or 4K that need high end cards 1080+ and that’s a small market to begin with. And then of that already small market only a smaller percentage will have the stomach to pay nvidias extortionate prices for 2080+ class cards.

      I mean I imagine I’m a perfect example… I’m looking for an UWD or 4K monitor upgrade but I’m not willing to spend the money nvidia is asking for the performance improvement they are giving at those resolutions. So I’ll wait for either a substantial price reduction on the 2080/Ti line (not happening) or i’ll wait another year for 7nm where if I’m going to pay the money, at least I’ll get uncompromised performance for UWD/4K resolution. And with any luck perhaps AMD will have a high end competitor in another 12 months time (though i’m not holding my breathe)

        • NovusBogus
        • 1 year ago

        It might have worked if real time raytracing had a fully baked and widely accepted software base backing it up, just waiting on drop-in hardware. But that isn’t the case; adoption and the implementation effectiveness thereof is a big question mark so potential customers are going to evaluate today’s card with today’s games, which puts RTX at a slight disadvantage even before factoring in the price.

          • NoOne ButMe
          • 1 year ago

          but RTX just works!

          That’s why games don’t have it ready 2+ months on, now that the Windows Update to support it is live!

          • Laykun
          • 1 year ago

          I mean, this is just a repeat of when shaders were introduced with the GeForce 3 or when the GeForce 256 came with hardware T&L. It’s ALWAYS going to be a chicken and egg problem and hardware makers have always been the one to push the envelope. RTX has great theoretical application and really is a good way forward, but just like with T&L and shaders the software just has to play catchup, but those who do eventually catch up will have a competitive edge.

          If you ever had to live with a GeForce 4 MX during the days of the 4 and FX series you’ll know the massive leap ahead it was and the pain of being stuck behind.

            • Spunjji
            • 1 year ago

            That’s not a great argument. I owned the GeForce 256 (the DDR, to be precise) and its performance in existing software was excellent, more than justifying the asking price. It only got better as software support came out. Of course by then you could buy a GeForce 2 for still more performance, but I never felt shafted by buying in early because I got the best the current gen had to offer in the interim period.

            So, it’s not “just like” that, because in this instance Nvidia have decided to pin the perf/$ equation. You pay a lot more for slightly better performance and a feature that may not even be worth having. In reality all indications are that this is less like T&L and more like tessellation, where AMD’s first attempts at the feature were not capable of providing the necessary performance for the feature to be utilized properly.

        • Kretschmer
        • 1 year ago

        2560x1440p appreciates a 1080Ti if you want smooth strobing.

        And looking at the panel pipeline, 1440p is going to have w “3440” in front of it most of the time (granted, 4K gaming is still an idiot trap).

          • Waco
          • 1 year ago

          4K gaming is pretty awesome even on an older card if you aren’t going crazy on settings. I game at 4K with an old Titan Xm…

            • Spunjji
            • 1 year ago

            I had a fair bit of fun in Fallout 4 and GTA 5 at 4k with the GTX 980. It was interesting deciding between 1440p at higher settings and 4K with moderate settings. Some games the resolution seems to be worth it, although that’s assuming you’re the type of player that find ~40fps to be fine.

            • Waco
            • 1 year ago

            I turn down AA and disable motion blur at 4K and that usually brings things well into the playable range even with modern titles.

            I notice the reduction in sharpness far more than the slightly crappier shadows, textures, and aliasing.

          • swaaye
          • 1 year ago

          Yeah I sometimes run the 1080 Ti at 1440p. There are lots of games that bring the frame rate down to an irritating level at 4K. Even some older games, like Evil Within. Sometimes I’d rather reduce pixel count and instead kick the quality settings up while maintaining high frame rate.

          I feel like we’d really need something like >2x 1080 Ti performance for ideal current day 4K gaming. Nevermind the future.

      • Krogoth
      • 1 year ago

      Nvidia is about to do a price war in response to RX 590 and AMD stealthy reducing prices on Vega 56/64.

      Expect significant discounts on existing GP104, GP106 SKUs plus “1060Ti” on this upcoming holiday season into January 2019.

      • derFunkenstein
      • 1 year ago

      [quote<]RX 480 8GB[/quote<] What year is it?

        • rudimentary_lathe
        • 1 year ago

        Meh, 480, 580, not much difference. The point still stands.

          • Srsly_Bro
          • 1 year ago

          But did you know the 580 existed when you made your post or are you really stuck in the year of the 480?

            • rudimentary_lathe
            • 1 year ago

            Asking the important questions 🙂

            I did indeed know the 580 existed when I made my post. But I personally own a 480, so I have a good excuse for the slip.

            • Srsly_Bro
            • 1 year ago

            All is forgiven, my brother.

    • psuedonymous
    • 1 year ago

    [quote<]For the fourth quarter of Nvidia notes that "excess channel inventory of mid-range Pascal products" will weigh on its fourth-quarter results, although it expects that excess inventory to clear out of retail and e-tail in "one to two quarters." Nvidia says it intends to ship no "meaningful" numbers of mid-range Pascal cards into the channel for the fourth quarter to allow demand to draw down those stocks.[/quote<] Too translate, before the inevitable FUD: Nvidia are not holding onto ”a bunch of unsold Pascal dies" as is often mistakenly claimed. Instead AIBs have a bunch of competed cards in inventory. That Nvidia do not intend to ship any further mid-range dies to AIBs may indicate a Q4 launch for a new midrange lineup, or there may be sufficient stock that that would happen in Q1 (though Q4 makes sense for the holiday sales).

      • NoOne ButMe
      • 1 year ago

      Yes, Nvidia got stuck with a ton of Pascal dice from mining. Now they’re mostly in the hands of AIBs.

      • UberGerbil
      • 1 year ago

      If they aren’t already shipping chips to the board vendors, it’s a little late to spike holiday demand. Black Friday has become most of November and is already happening. I realize we’re not talking about little Timmy reading a review of a new GeForce, writing a letter to Santa, and his parents going out to a B&M store to look on the shelves for the thing, but there’s still a pipeline that needs to get primed and filled and that takes time.

      At some point nVidia will decide they need the hype and revenue from a new generation more than they need the goodwill from sheltering their board partners from getting Osborned, but it sounds like they’re willing to wait at least a quarter on that.

      This is what it looks like when a bubble pops.

        • psuedonymous
        • 1 year ago

        [quote<] This is what it looks like when a bubble pops.[/quote<] The crypto bubble popped damn near a year ago now. Any pricing pressure (upwards from demand or downwards from oversupply) will have had their effects months ago.

          • Krogoth
          • 1 year ago

          We have already seen the effects with the burst when prices went back to launch MSRP levels. They are going to be going be going below that threshold in the next quarter or two.

            • psuedonymous
            • 1 year ago

            The drop to just above MSRP at the start of this year [b<]was[/b<] the 'crash'. Since then, pricing pressure has been from historically high DRAM prices and overcapacity at TSMC. That's why the price floor has remained so high.

            • Krogoth
            • 1 year ago

            Actually that’s only first crash but it was still profitable to mine with GPUs so prices were still massively inflated. Nvidia most likely kept putting in massive orders to their vendors in an attempt to milk the market for its worth. The second crash in June/July 2018 is what caused the crypto-currency bubble to finally deflate and it also when prices started to return back to normal. Nvidia’s card partners got stuck excessive stock on some SKUs (namely 1060s and 1070s).

            The real reason why they never went below launch MSRP? Nvidia had no serious competition in the performance tier GPU market and their shareholders wanted to keep up the healthy profit margins. They are only starting to cave into the pressure from board partners and indirectly etailers/B&Ms who want to reduce prices to move old inventory because lesser next-generation SKUS are coming soon. They don’t want to be stuck with “osborned” products.

            Nvidia will mostly likely do price war to coincide with “1060Ti” launch to spoil RX 590 and AMD RTG’s attempt to snag any potential holiday sales before lesser Turings and Navi come around. Nvidia wanted to AMD RTG make the first move in the price war.

            • psuedonymous
            • 1 year ago

            [quote<]Nvidia most likely kept putting in massive orders to their vendors in an attempt to milk the market for its worth. [/quote<]AIBs order chips from Nvidia, not vice versa.

            • NoOne ButMe
            • 1 year ago

            If NVidia tells you “you have to order 1000 GTX 1060 and 200 GTX 1070 GPUs to get 100 2070 and 100 2080 GPUs” what do you think happens?

            Not saying Nvidia is doing exactly that, but Nvidia is likely favoring shipments to vendors who are willing to buy Pascal cards.

            Either way, NVidia has a massively lower projection, and this 3Q result was still a massive gain over their previous 3Q. 4Q has been the largest by revenue for the last 4 fiscal years. Usually a little higher than 3Q.

            Now, what could make 3Q result very high, while negative impacting 4Q results?

            actually looking closer, their sales this year have been very level. So while expecting a large jump in 4Q might be unreasonable, one would certainly not expect a giant drop.

            • Srsly_Bro
            • 1 year ago

            Blah blah blah fake hypothetical to prove my fake point blah blah blah with no insider info blah blah blah

            You sound like a guy i went to school with who would attempt to prove any point by use of hypothetical arguments.

            Reminds me of ancient aliens, where f*** all to finding an answer and just saying aliens.

            You said:

            Not saying Nvidia is doing exactly that, but Nvidia is likely favoring shipments to vendors who are willing to buy Pascal cards.

            Sounds like:

            Not saying it was aliens… But it was aliens

            Get the point?

            You’re welcome,

            Srsly_bro

            • Krogoth
            • 1 year ago

            Actually, it is a contract between Nvidia and AIBs.

            Nvidia and AMD RTG don’t have their own card assembly plants so they need partners to assemble cards. They order silicon from fabs (TSMC/GF) but they need card partners to build the cards. They split up cost and profit margins.

            Nvidia branded cards are actually assembled by PNY while AMD RTG branded stuff is assembled by Sapphire.

            • psuedonymous
            • 1 year ago

            [quote<]Actually, it is a contract between Nvidia and AIBs.[/quote<] It's a heck of a lot more contracts than that, between a lot more partners. An AIB wants to build x thousand 1080Tis. They order x thousand dies from Nvidia (who then order them from TSMC), x thousand memory packages from SK Hynix/Samsung/etc (or rather, fro ma supplier who orders from them), x thousand PCBs from a PCB house, hassive numbers of sundry components from other suppleirs (passives, MOSFETs, power controllers, etc), and then contracts with an OEM (e.g. Manli) to actually assemble and test those cards. [quote<]They order silicon from fabs (TSMC/GF) but they need card partners to build the cards. They split up cost and profit margins.[/quote<] The point is that Nvidia are not going to AIBs "build us X thousand cards" and then demanding a percentage of the revenue. That's not even vaguely close to how things work. [quote<]Nvidia branded cards are actually assembled by PNY while AMD RTG branded stuff is assembled by Sapphire.[/quote<]The same is true for everyone else. An 'EVGA' card will not be built at a factory with an EVGA logo on the side, for example.

      • HERETIC
      • 1 year ago

      Nvidia are not holding onto ”a bunch of unsold Pascal dies”

      So do all the stories of large quantities of dies being returned to NV, now
      become “FAKE” news?????

Pin It on Pinterest

Share This