NVIDIA Corporation
NVIDIA is best known as a producer of GPUs (Graphics Processing Unit) for computers and gaming consoles. Their GPUs are also being used in sophisticated computer systems for the purposes of complicated information processing, deep learning and artificial intelligence (AI). They see a significant opportunity in autonomous driving in providing computational solutions for AI rather than just sensor or detection solutions. It’s not cheap but their long-term growth opportunity explains their valuations. Recent results have been very strong, partly driven by very strong sales to Nintendo’s new platform. Due to the nature of product ramps, these recent results might be unsustainable in the near term and could result in disappointing results and pressure on the stock. Why would I buy in light of this possibility? It might not happen or it might be offset by growth in the rest of their business. In short, I bought a little bit hoping it will go down but in case it does not. I added more as the stock weakened along with most semiconductor stocks and now the company is licking its wounds due to an inventory correction in the channel due to excess inventory left over from the crypto currency boom. The challenge now is determining how long this lasts and if past results benefited from this boom more than management realizes.
Ests: 3Q19 4Q19 FY19 FY20
Revs 3.2b 3.4b 13b 14.7b
EPS 1.71 1.81 7.3 7.95
P=202.39, div=0.64, yield=0.2%, cash/sh=9.20, TTM EPS 7.59, ex cash P/E=26X, FY20 P/E=25X
Mid Quarter Update
Jan 28, 2019 – NVDA warns on 4Q19
Expects Revs at 2.2bn vs previous guidance 2.7bn
Excess mid-range channel inventory reduced largely as mgmt. expected however deteriorating macro conditions “particularly in China”, is impacting gaming. High end GPU sales were also lower than expected due to higher prices and consumers waiting for games to adopt some of the technology like ray tracing.
Datacenter was also below mgmt. expectations as a number of anticipated deals did not close as customers are getting more cautious (from an inventory perspective or actual deployment?)
Also expects 120m in charges for excess DRAM and other components to impact GM
Implies EPS ~0.76 including .17 hit from DRAM charges (est. was 1.40). As the company was putting up huge beats last year, this huge miss demonstrates the leverage both ways.
When they reported their 3Q results and gave their 4Q guidance, I was optimistic that the issue in gaming from crypto-related sales would not not spread to the other parts of their business and I was wrong and too early to be buying the shares on that expectation. At best they are facing an inventory adjustment across the rest of their business and at worst they are facing weaker end-demand that could last longer. As such, I'm re-evaluating whether NVDA should be among my top holdings especially if the macro environment continues to deteriorate.
3Q19 1.84 vs 1.33, +38% est 1.71 down from 1.82
Nov 15, 2018, P=202.39, TTM EPS=7.59, ex cash P/E=26X, FY20 P/E=25X
Revs +21% to 3.18bn, GM 61% vs 59.7%, OM 38% vs 38.1%, OpInc +20%
Gaming +13% to 1.76bn, Professional Visualization +28% to 305m, Datacenter +58% to 792m, Auto +19% to 172m, OEM and IP -23% to 148m
While Datacenter was +58% vs +26% for Intel, it was only +4% q/q while Intel was +11% q/q
Tegra -3% y/y and -13% q/q seasonal decline for gaming consoles.
Inventories +30% to 1.42bn q/q +65% y/y due to prodn ramp of Turing which started shipping later in the quarter, could be a red flag that past demand was running hotter than sustainable but much of the increase is raw materials and WIP, finished goods is 524m, +121q/q and y/y
AR has increased 557m q/q and 1bn y/y to 2.2b also likely due to Turing late launch. I’d expect to see both work down over the next couple of quarters otherwise could be a red flag. Last year inventories were below 1 quarter of COGS and now they are slightly higher than 1 quarter so it’s not necessarily a red flag yet (1 customer is 14% of A/R, not out of whack with history)
Softness reflects excess inventory in the channel post crypto-currency boom (not surprising, this has been communicated by AMD and general supply chain correction by other semi companies, mgmt. expects adjustment to take 1-2 quarters. Guidance assumes no meaningful shipments of mid-range Pascal GPUs in 4Q so channel can adjust and minimal Tegra shipments for gaming consoles due to seasonality, other markets unaffected.
Raised dividend 7% to 0.64, expects to return additional $3bn in FY2020 which could being in 4Q19
4Q Guidance: Revs 2.70bn +/1 2% vs 2.91, -7% y/y and -15% q/q, est 3.4bn, GM 62.5%, OpEx 755m, Other income 21m, tax rate 8%, implies EPS 1.40, -19% y/y, est 1.81.
Mgmt misjudged the channel which was slower to lower prices than mgmt. expected and consumers were waiting for price declines, prices now normalized.
Midrange has been ~1/3 of gaming over past 8 quarters or so, reasonable to assume approx 30% q/q decline in gaming and rest of business to grow q/q, this would imply rest of business ~+25% y/y (decelerating from recent quarters). Encouraging that management said it’s been pretty consistent around 1/3, mgmt. feels Pascal still strongly positioned in midrange.
High-end Turing had a great quarter according to CEO, solid transition from high-end Pascal to Turing
While not entirely surprising, results are disappointing although the stock has corrected to anticipate weak results and guidance, momentum investors will get out and cause the stock to overshoot, fundamental investors will try to gauge normalized earnings power, and many will wait to see the proof in the pudding for stabilization. Possibly the more disappointing aspect is a year ago I was cautious of inventory correction but continued strong results assuaged my concerns, thankfully issue appears limited to cryptocurrency business rather than the business at large.
While they were posting strong growth, valuation looked reasonable. The question is whether this is a short-term setback or whether sales from the last few quarters were unsustainably hot not only skewing the level of sales and earnings but also the growth rate.
If 4Q is a kitchen sink quarter and there are no sales of Pascal and minimal sales for gaming consoles, and the rest of the business is growing, this would be a reset and annualizing $1.40 in earnings might be a reasonable approach. So if earnings power is $5.60 and growing, $160 would put the stock at 28X its earnings power which wouldn’t be horrible as long as the company puts up solid growth after the reset although it is not yet determined if investors will accept this multiple given the uncertainty. This will likely take time to unfold and there certainly will be an abundance of impatient investors. It’s unlikely though that the company will return to recent earnings levels in the near future.
Looking back at my trades, I trimmed twice in March and was hasty in repurchasing those shares and a bit extra in May. I then added twice more as the stock weakened. I still believe in the long-term opportunity and I know that is not always linear but it’s clear I boosted my weighting too early (currently 2.61% of my portfolio). I accepted that management had a good handle on channel inventory (wrong) and that Datacenter was still going strong (true but pales in comparison to the size of the miss in gaming) and that automotive still presents a big opportunity (true but not immediately).
2Q19 1.94 vs 1.01, +92%, est 1.66 up from 1.48
August 16, 2018, P=257.44, TTM EPS=7.05, ex cash P/E=35X, FY20 P/E=31X
Revs +40% to 3.12bn (est 3.1), GM 63.5% vs 58.6%, OM 41.3% vs 34.7%, OpInc +90%
Gaming +52% to 1.8bn, Professional Visualization +20% to 281m, Datacenter +83% to 760m, Automotive +13% to 161m, OEM and IP -54% to 116m due to lower crypto
Their guidance had anticipated 100m in crypto but came in at 18m, expect 0 going forward.
3Q Guidance: Revs 3.25m (+23%, est 3.3bn), GM 62.8%, OpEx 730m, tax rate 9%, implies EPS 1.94 (+46%, est 1.82), y/y growth could decelerate for a few quarters as they lap higher crypto sales although I’m pleased that so far crypto hasn’t turned out to be a bigger part hidden in gaming contributing to significant weakness, growth still looks solid.
Results are solid despite crypto going almost to 0, guidance is fine although the stock is weak after hours because revenue outlook is 1.5% below estimates (likely because analysts were expecting some crypto revs), EPS guidance still solid and ahead of estimates.
Just launched “the biggest leap in GPU architecture in over a decade for photorealistic rendering” – Turing – and should open up Professional Visualization in instances where they couldn’t use GPUs. Also expect to drive gaming graphics., can also be used for high throughput deep learning inferencing for data centres.
1Q19 2.05 vs .85, est 1.45 up from .98
Revs +66% to 3.2bn (est 2.9bn), GM 64.7% vs 59.6%, OM 44.5% vs 32.9%, OpInc +124%
Gaming +68% to 1.72bn, Professional Visualization +22% to 251m, Datacenter +71% to 701m, Automotive +4% to 145m, OEM and IP +148% to 387m
Supply was tight earlier in the quarter but now easing, channel prices for GPUs beginning to normalize
Crypto currency business was strong ~289m, fulfilled most with crypto-specific GPUs and included in OEM segment, expect 2Q crypto revs to be 33% of 1Q
2Q Guidance: Revs 3.1bn (est 2.95), GM 63.5%, OpEx 685m, tax rate 11%, implies EPS 1.83, est 1.48
Outstanding results and guidance, unlike in recent quarters where estimates were for moderate growth, EPS was expected to grow 71% while they grew 141% (still massive beat). Datacentre still a huge opportunity, while growing significantly faster than Intel’s Datacenter business y/y, the growth was smaller in dollars. However q/q Intel’s datacenter business contracted 400m while NVDA grew almost 100m q/q.
4Q18 1.72 vs 1.13, est 1.16 up from .97, FY18 EPS 4.92 vs 3.06
Revs +34% to 2.91bn, GM 62.1% vs 60.2%, OM 41.3% vs 37.2%, OpInc +49%
Gaming +29% to 1.7bn, Professional Visualization +13% to 254m, Datacenter +105% to 606m, Automotive +3% to 132m, OEM and IP +2% to 180m
Crypto revs higher than in 3Q but amount not disclosed, does not appear to be a material driver of revs, channel inventories of gaming GPUs (6-8 weeks) lower than historical due to high demand.
Automotive seeing declines in infotainment commoditization with growth in next gen AI cockpit systems
1Q Guidance: Revs 2.9bn +/- 2% (rev acceleration +49%, est 2.47), GM 63% +/- 50bps, OpEx 645m, tax rate 12%, implies EPS ~1.66 (+95%, est 0.98)
Again, NVDA crushed it delivering EPS growth of 52% when analyst estimates were for +2.6% when 3 months ago estimates were for a DECLINE of 14%, 1Q Guidance is for rev growth to accelerate and EPS growth to be +95% vs expected growth of +15%. The market cheered Intel’s datacenter business being +20% and NVDA was +105% (on top of +205% last year), Intel’s 4Q18 datacenter revenue was 9X bigger than NVDA’s, datacenter opportunity alone is massive then there’s the nascent autonomous driving, AI, and machine learning. NVDA’s FY18 datacenter revenue was 1.93bn vs 830m in FY17 and 339m in FY16.
Potential risk is in high-demand environment, end users buy in advance of need, depleting channel inventory and overstating near-term demand trends however the dynamic revenue growth and ability to exceed guidance might be indicators of limited constraints.
3Q18 1.33 vs .94, est .94 up from .79
Revs +32% to 2.64bn, GM 59.7% vs 59.2%, OM 38.1% vs 35.3%, OpInc +42%
Growth across all platforms, Gaming +25%, Professional Visualization +15%, Datacenter +109% (+20% q/q after last quarter’s perceived “softness”), Automotive +13%, OEM and IP +3%
4Q Guidance: Revs 2.65 +/- 2% (est 2.43bn), GM 60% +/- 50bps, implies EPS 1.30 (est .97)
Intends to return 1.25bn to shareholders in FY19, same as FY18 goal.
Great quarter, to simply sum it up, analysts were expecting NO EPS growth y/y and we got 60%. If that's not a blowout, I don't know what is.
2Q18 1.01 vs .53, est .70
Revs +56% to 2.23bn, GM 58.6% vs 58.1%, OM 34.7% vs 26.8%
Shipped units of the new Volta-based V100 accelerator, provides 10X the deep learning power from the previous GPU which is 1 year old.
3Q Guidance: 2.35bn (est 2.13), GM 58.8%, implies EPS 1.06 (est .79)
Revs and EPS came well ahead of estimates, so does guidance. Stock selling off 6% after hours after a weak day but the strong has been very strong recently (+60% since 1Q18 report) so a breather is not out of the ordinary (taking back 1 month’s of gains so far) especially as investors try to rationalize 42X trailing EPS with EPS above estimates for the fiscal year and next year. Some investors appear to focus on Datacentre being only +2% q/q (co says transition quarter, it was +178% y/y), in 2Q17 it was +5.6% q/q and in 2Q16 it was -18% q/q. Sequential results (q/q) can send misleading signals unless it’s a true cyclical pattern. If the 2Q16 -18% was to send a negative warning, 2Q18 vs 2Q16 was 416m vs 72m or +478% in just 2 years. That said, although guidance was ahead of estimates, it’s for Revs +17% y/y or +5% q/q so a deceleration of growth not seen for 6 quarters so certainly some investors will be forming conclusions whether this signals a coming peak in its business or a pause in growth.. the coming applications and implementation of AI support long-term growth.
1Q18 .85 vs .46, est .67
Revs +48% to 1.94bn, GM 59.6% vs 58.6%, OM 32.9% vs 24.7%
Gaming +49% to 1bn, Professional Visualization +8% to 205m, Datacenter +186% to 409m, Auto +24% to 140m, OEM and IP -10% to 156m, IP from Intel was 43m, the remainder of that agreement
2Q Guidance: Revs 1.95bn (est 1.9), GM 58.6%, implies EPS~.79, est .62
Solid results, difficult to tell if sales into Nintendo are at higher than normal levels but overall business is performing well and valuation does not appear to be excessive.
4Q 17 1.13 vs .52, est .83, FY EPS 3.06 vs 1.67
Revs +55% to 2.17bn, GM 60.2% vs 57.2%
FY Revs +38%
1Q 18 1.9bn +/- 2%, EPS 0.77, vs .46, est .64