Tech Stock Thoughts #11: GPU Prices, Tesla/Bitcoin, Qualcomm, Twitter, Facebook
Some thoughts on gaming GPU demand, Tesla’s Bitcoin purchase, Qualcomm’s modem and RF chip announcements, Cisco and Twitter’s earnings, Facebook’s reported smartwatch plans, and the recent selloff in high-multiple tech stocks.
High-End Gaming GPU Prices Are Higher Than Ever — Unless a seller gives it a lower listing price, graphics cards based on Nvidia’s flagship GeForce RTX 3090 GPU ($1,499 MSRP) are now typically selling for $2,600-$2,800 on eBay, up from around $2,000 in December. And RTX 3080 graphics cards ($699 MSRP) are going for around $2,000, up from $1,200-$1,300.
Likewise, cards based on AMD’s flagship Radeon RX 6900 XT GPU ($999 MSRP) are selling for $1,600-$1,800, up from $1,350-$1,400 in December. And RX 6800 XT graphics cards ($649 MSRP) are selling for $1,300-$1,500, up from $1,050-$1,100.
The PS5 and Xbox Series X are also still fetching decent aftermarket premiums, albeit ones that are similar to what they saw 6 weeks ago. The standard PS5 tends to sell for $350-$400 more than its $499 MSRP, while the Series X tends to go for $200-$250 more than its $499 MSRP (that price delta probably says a bit about which console is hotter right now).
There’s still clearly a ton of demand for high-end gaming hardware in general, and capacity shortages for advanced TSMC and Samsung processes aren’t helping with the supply/demand balance for either graphics cards or next-gen consoles. But it’s nonetheless telling that graphics cards premiums have kept rising, while console premiums have remained fairly steady. It looks as if Ethereum’s giant run-up over the last couple of months has added fuel to the fire.
Related: To satiate demand for less powerful GPUs in this environment, Nvidia is re-releasing its Turing-architecture RTX 2060 and Pascal-architecture GTX 1050 Ti GPUs to graphics card makers.
Qualcomm Unveils a New Flagship 5G Modem and RF Front-End Chips — Getting the most attention here: Qualcomm’s Snapdragon X65 modem supports the 3GPP Release 16 5G spec and has a 10Gbps peak downlink speed — 2.5Gbps higher than its predecessor, the Snapdragon X60, and 2.3Gbps higher than MediaTek’s recently-announced M80 modem. But there are a ton of other improvements as well. Among them: Support for software-based radio upgrades, improved power amplifier envelope tracking, and an antenna-tuning solution that leverages AI/ML to detect when a user’s hand is blocking an antenna.
Qualcomm’s aggressive modem and RF R&D spending looms large here, as does its status as the only company offering end-to-end 5G radio platforms (covering modems, sub-6GHz RF chips/modules and mmWave antenna modules). MediaTek is still a tough mid-range modem/app processor rival and Skyworks, Qorvo and Broadcom are all well-represented in sub-6GHz RF, but Qualcomm does look poised to maintain a dominant position in the high-end 5G modem market and add to its recent RF share gains.
Tesla Buys $1.5B Worth of Bitcoin — A few quick thoughts about this:
- The purchase is equal to less than a tenth of Tesla’s Q4-ending cash balance, not to mention just ~0.2% of its market cap. As such, it’s arguably something of a branding/marketing exercise for a company that still doesn’t spend a penny on advertising. And assuming the bottom doesn’t fall out of Bitcoin’s price (a risk, admittedly), it might be money well-spent in that respect.
- Execs and PR/IR departments at other growth-stage companies are probably taking notes. While I’m skeptical of the Apple/Bitcoin speculation, given Apple’s approaches to both branding and capital allocation, it’s not hard to imagine many tech firms with large retail followings seeing Bitcoin purchases as a way to further endear themselves with Robinhooders.
- I wrote a few weeks ago that I wasn’t sold on Bitcoin’s January correction being the start of another crash. With Bitcoin having soundly taken out its old high following Tesla’s announcement and surged past $50K on Wednesday, I think the stage could now be set for one final blow-off rally, after which we could finally see a lengthier correction/cooldown period.
Disclosure: Long Bitcoin (have held since 2018 outside of a couple of trades when the chart went parabolic in 2019, am open to temporarily selling again if there’s another blow-off rally). Short Tesla (took a position a couple weeks ago, just as a valuation-driven trade).
The Push for Localized Chip Supply Chains Hits a Tipping Point — Over the last week:
- An open letter — signed by basically a who’s who of U.S. chip industry CEOs — calling for “substantial funding for incentives for semiconductor manufacturing, in the form of grants and/or tax credits, and for basic and applied semiconductor research” was sent to President Biden.
- The Biden Administration said it plans to sign an executive order that’s both meant to help deal with current chip shortages (affecting everything from cars to notebooks to gaming hardware) and kick off a “comprehensive review of supply chains for critical goods.”
- Bloomberg reported the EU is interested in seeing an advanced fab built on the continent, and that TSMC or Samsung could be involved.
All of this follows TSMC’s decision to build a 5nm fab in Phoenix that’s promised to start production in 2024, as well as a report that Samsung is thinking about building a 3nm fab in Austin that could go live as soon as 2023 (an aggressive timetable, admittedly).
If there were any lingering doubts that a major effort would be undertaken to reduce the dependence of U.S. and European chip developers on overseas leading-edge fab capacity, they should be gone by now. And that’s of course pretty good news for chip equipment makers. It’s also good news for TSMC, Samsung and Intel to the extent that their future capex and process R&D is partially subsidized, and it might even be good news for fabless chip developers to the extent that more fab capacity is available to them.
Chip stocks rallied sharply on Thursday and Friday thanks to the news, with equipment makers generally leading the way (for the moment, my post-earnings suggestion that investors wait for a pullback to buy KLA isn’t looking great). With so many chip stocks already trading near their highs, markets might’ve gotten a little carried away, particularly since some of the new U.S. and European fabs that are built will inevitably take away from capex that would otherwise occur in Taiwan and South Korea. But the news does add one more bullet point to the bull case for foundries and equipment makers, and it might just make the industry’s next capex down-cycle softer than it otherwise would be.
Cisco’s Earnings — January quarter results slightly topped expectations, as did April quarter sales guidance. But revenue was flat, product orders were up just 1% and Cisco is only guiding for 3.5%-5.5% April quarter growth in spite of easy comps. Even with shares only going for 15x FY21E EPS into the print, it’s hard to blame markets for wanting a little more.
Also, in what’s yet another sign of weak on-premise enterprise IT spend amid accelerating cloud migrations and weak demand from COVID-impacted verticals, Cisco’s enterprise product orders were down 9%. Telco orders were also down (this fits with what Juniper reported seeing), and Cisco’s Applications segment revenue was flat, with “strong double-digit” Webex growth (likely still far below what Zoom and Teams are seeing) offset by declines for unified communications (RingCentral has been taking share here) and telepresence gear.
There were admittedly some bright spots. Security revenue and public sector product orders were both up 10%. And orders from hyperscalers rose by a triple-digit percentage and now account for a quarter of service provider orders, as Cisco continues seeing a payoff from its late-2019 decision to start selling switching/routing processors and optics to hyperscalers a la carte (going forward, Cisco’s pending acquisition of optical component/module vendor Acacia Communications should give these efforts an additional boost). But competitive and secular headwinds are still making it hard for Cisco to register anything more than modest top-line growth.
(Cisco’s earnings report, earnings slides)
Twitter’s Earnings — Sometimes, having an up-close look at everything that a business is doing wrong can blind you to how little it actually needs to do right to meet or beat the market’s expectations, particularly when there are also known near-term headwinds in play. Twitter, which soared post-earnings in spite of missing its Q4 mDAU consensus and only guiding Q1 revenue and mDAUs in-line, provided me with a reminder of this.
As a long-time user, I could probably write several columns about everything I think is wrong with Twitter — from its issues with content discovery; to its shaky ad targeting; to how it makes regular users feel ignored; to how it fosters the creation of (and incentivizes pandering to) cliques, echo chambers and outrage mobs; to how it depresses engagement and makes even replies and likes performative by making followers aware of all user activity save for DMs; to its lack of monetization options for users with large followings (changing soon, maybe); to how — in spite of widespread interest in such features, as shown by the popularity of screenshots and tweet threads — it still makes it far more difficult than it should be to share or have discussions around lengthier commentary and arguments.
But in spite of all that, Twitter remains a very useful platform for discovering facts and insights on topics of interest, getting almost-immediate dispatches, commentary and one-liners related to major news events, and hearing from (and at times, engaging with) prominent figures in various fields. And when your DAU count is less than 1/9th as large as Facebook’s and even 25% smaller than Snapchat’s, you only need to do so much right to grow, especially in the midst of a pandemic.
As a result, even with Donald Trump’s exit from both Twitter and the White House acting as usage headwinds in the U.S., Twitter guided for mDAUs to be up ~4% Q/Q and ~20% Y/Y in Q1. And with the help of both a booming online ad market and efforts to revamp a long-struggling mobile app install ad business, Twitter’s Q4 revenue topped consensus by more than 10%, with Y/Y growth improving to 28% from Q3’s 14%.
This doesn’t necessarily mean that Twitter, which now sports a $57B market cap, deserved to pop as much as it did post-earnings. Q1 revenue guidance was just in-line; Twitter didn’t offer specifics about how U.S. mDAUs — just 19% of total mDAUs, but responsible for ~55% of all ad revenue — are trending in Q1; and the company did say its full-year guidance for revenue growth to exceed cost/expense growth of “25% or more” assumes both macro improvement and only a “modest impact” from iOS 14’s privacy policies. And while it’s encouraging to see Twitter explore subscriptions (again!) and test out a Clubhouse alternative, the company’s execution with new features/services has often left much to be desired. Just consider how quickly Fleets have become an afterthought for most users.
But still, it is fair to say that Twitter’s outlook was on the whole better than what I and others thought it might deliver. Lesson learned.
(Twitter’s earnings report, shareholder letter)
Apple Reportedly Teams with TSMC on AR Headset Displays — According to Nikkei, Apple is looking to use micro OLEDs printed on silicon wafers (rather than glass substrates, as OLEDs typically are) to act as the displays for its AR glasses (reportedly due in 2023, per other reports).
Micro OLEDs present manufacturing challenges (Nikkei’s sources say it’ll take several years to reach mass-production) and will undoubtedly be costlier than standard OLEDs, but also provide major advantages in terms of size, thickness and power draw. Just one more sign that Apple, which is also prepping a low-volume AR/VR headset for 2022 that will feature dual 8K displays, is taking a first principles approach to crafting its AR and VR headsets.
NBCUniversal Reportedly Mulls Peacock Options as Disney+ Chugs Along — According to The Information, only 11.3M of the 33M people signed up for Comcast/NBCU’s Peacock service qualify as “monthly active ad-supported accounts.” In addition, only 4% of Peacock’s sign-ups are said to involve its ad-free, $10/month, Premium plan.
All of this has led NBCU to consider its options for Peacock. The Information’s sources say a bundling deal has been pitched to ViacomCBS, and the site also speculates (noting that NBCU chief Jeff Shell once told colleagues that he supported such a move) that NBCU could merge with HBO parent WarnerMedia (owned by AT&T) to obtain more scale.
Consider me unsurprised (see 2021 tech prediction #8). Streaming is very much a scale and mindshare game, and neither NBCU nor ViacomCBS has enough of these things relative to Netflix, Amazon, Disney and to a lesser extent WarnerMedia. A major overhaul of their streaming strategies — either a merger with a larger player, or a shift to licensing content to the highest bidder — might just be a matter of time.
Meanwhile, Disney+ continues adding subs at a brisk pace ahead of its planned March price hike. Disney+ had 94.9M paid subs as of Jan. 2, up from 86.8M a month earlier and just 26.5M a year earlier. Even if one backs out subs for the Indian Disney+ Hotstar service (said to account for ~30% of the subscriber base), those are pretty remarkable numbers. And with a price hike and giant content investments looming, Disney looks well-positioned to copy from Netflix’s playbook in the coming years.
Facebook Reportedly Plans to Launch an Android Smartwatch in 2022 — This story, which also comes via The Information, was a little surprising at first glance. Smartwatches, after all, are hardly an ideal interface for browsing social feeds, and while they can be useful for quickly seeing messages, they’re also not an ideal interface for composing them.
Also, the market is pretty competitive, with Apple quite dominant on the high-end and possessing a slew of competitive strengths. And in spite of speculation to the contrary, I’m pretty skeptical that Facebook cares to invite the backlash that would ensue if it used health/fitness data obtained from a smartwatch for ad targeting (Google, it’s worth noting, promised not to use data produced by Fitbit’s hardware for such purposes in order to obtain antitrust approval for the Fitbit acquisition).
Thinking about it, however, a couple of plausible reasons for a smartwatch effort came to mind:
- With Mark Zuckerberg having declared in 2019 that he expects “future versions of Messenger and WhatsApp to become the main ways people communicate on the Facebook network,” Facebook might feel a need to have a meaningful presence on any device that’s often used for messaging.
- With Facebook currently working on AR glasses and having bought neural interface startup CTRL-Labs — it’s working on a wristband that aims to let people control devices with their thoughts — Facebook might see a smartwatch platform as complementary to other wearable platforms it’s developing.
Facebook’s track record with hardware is somewhat mixed, and based on what The Information is reporting, I wouldn’t count on its first smartwatch selling in huge volumes. But the company does tend to think long-term with its hardware projects, and its history with Oculus and Portal devices point to a knack for iterating on niche products until they can appeal to a broader set of consumers.
Disclosure: Long Facebook (took a position in early January).
High-Multiple Tech Stocks Are Coming Under Pressure — Over the last week, various high-multiple clean energy, 3D printing, SaaS and WFH plays have meaningfully sold off even as the Nasdaq has largely held steady. The selling has been pretty orderly, but the losses have gradually piled up for quite a few Robinhood favorites.
In addition, a slew of high-multiple SaaS/growth software names — think companies such as Datadog, Cloudflare, RingCentral and Palantir — have sold off post-earnings over the last week, in spite of posting fairly strong Q4 reports that might’ve been good enough to drive additional gains three months ago. Combine this with the large post-earnings selloffs recently seen in companies such as Peloton and Unity Software, and it does look as if the bar is getting set higher now.
And all of this of course comes after the GameStop/Wallstreetbets debacle, which undoubtedly served as a wake-up call to many newer investors that buying frenzies fueled by herd behavior can and often do end badly.
High-multiple retail favorites have bounced back from corrections with a vengeance several times since last summer, and so in spite of the misgivings I have about the valuations of many of these companies, I’ll steer clear of predicting an imminent crash. But there has been a change in the mood lately.
