When it comes to running your computer, smartphone or gaming console you have the CPU (central processing units) and the GPU (graphic processing units). While there are lots of other components, these are the workhorses. The first basically completes all the tasks dictated by software, while the latter generates the images you see on the screen. With the maturing of PC and smartphone markets, and the slowing in advancement of chips speeds, CPU pricing has come under pressure, making it a less profitable space to be in. That’s a major reason for the surge in M&A in the chip space since the start of 2015 as companies scramble to make sure they’re part of the next wave of products. The latest rumored deal last week:
Qualcomm (QCOM) is in talks to acquire NXP Semiconductors (NXPI) for $30 billion... Since the start of 2015 there has been about $200 billion in semiconductor m&a as the PC & Smartphone markets mature, component suppliers are desperate to diversify into emerging technology products & platforms like the artificial intelligence (AI), augmented & virtual reality, Internet of Things (IoT), autonomous cars, drones etc.
Back in July, on his firm a16z’s July 10th podcast (Software Programs the World), famed VC Marc Andreessen highlighted the emergence of new platforms (generally distributed systems, platforms built across lots of chips), with his first example being Amazon’s AWS public cloud. But he was even more geeked up about the possibilities of Artificial Intelligence (machine learning).
And AI isn’t just about CPU, it eventually leads back to GPUs (not just gaming anymore). Basically a “new application for an old idea” where GPUs are being used for AI processes beside the CPU to “run virtual simulations much faster”. This is a bit above my pay-grade, but the main point is simple, PCs & smartphones will be lucky to ever grow units double digits ever again, and semiconductor companies and investors alike have been playing one of the most epic games of musical chairs positioning for the next really big thing, and everyone has a stake.
Most of the M&A focus so far has centered around the Internet of Things (IoT). And the premium to sales have been massive, with INTC’s $17bn bid for ALTR last year,and ADI’s recent $15bn bid for LLTC, both about 10x trailing 12 month sales of about $1.5bn, only outdone by Softbank’s $32bn bid for ARMH at a whopping 22x sales!! It is important to note though, the greater the annual sales the lower the premium that’s been paid
But what about the deals in GPU-land to get pole positioning for AI? The leader in GPUs is Nvidia (NVDA). Andreessen suggested that NVDA has “seemingly overnight become the leader in chips for AI”, going on to say that “every sharp AI entrepreneur that comes in (to their VC firm) is now building programs on NVDA chips”.
If this is news to you, it is not news to big investors, shares of NVDA are up 108% this year, up 175% from its February lows, and now trading at new all time highs.
The stock has a $36bn market cap ($3.3 billion in net cash), about 6x its expected f20176 sales . Oddly, consensus estimates only have eps growing 12% in f2018 (next year) on a 7% yoy sales increase. Either consensus is very wrong, and Andreessen’s observations are yet to be fully factored in for growth in AI chips, or the market is way over paying for takeout potential.
But there were a couple announcements from a couple stone cold playas (AMZN & MSFT), where AI will be a big part of most of what comes next for both, per Barron’s Tech Trader:
One bit of news was Amazon.com (AMZN) announcing this morning it will rent use of Nvidia’s graphics processing unit (GPU) chips in its Amazon Web Services, as what are called “P2 Instances,” boasting, “With up to 16 NVIDIA Tesla K80 GPUs, P2 instances are the most powerful GPU instances available in the cloud.”
The other bit of news was Microsoft (MSFT) yesterday saying it’s forming a new division, “Microsoft AI and and Research Group,” which will be staffed with more than 5,000 people brought together from across the company. The company also said it will be rolling out some Azure offerings with Nvidia chips.
Regardless, if you are kicking the tires now for NVDA’s opportunity away from gaming in AI, the ship has sailed. There are few who could, assuming anyone would (few expected Softabank as an entrant into the semi M&A space) pay 10x NVDA’s expected sales (close to $60 billion). Maybe 8x. Maybe. But if anything, with its stock where it is, and the strong balance sheet, NVDA is more likely to be an acquirer than an acquiree.
So What’s the Trade?
The next identifiable catalyst for NVDA will be fQ3 results the first week of Nov. The stock has reacted fairly well to earnings over the last 6 quarters, with an average next day rally of about 9%, seemingly making new highs in each instance.
Short dated options prices look downright cheap, with 30 day at the money realized volatility (the price of options – blue line below) just a hair above 30 day realized volatility (how much the stock has been moving – white below):[caption id="attachment_66793" align="aligncenter" width="600"] From Bloomberg[/caption]
This stock is right in the middle of huge transition in the space. That disruption has companies scrambling. We’ll be sure to circle back with some trade ideas as we get closer to NVDA’s next event. For those looking to position ahead of that the best bet at this point may be calendar spreads, selling an October expiration call or put versus buying in November to capture earnings. But even that comes with some risk as the strike chosen needs to have some room for any moves in October. For instance, the Oct/Nov 70 call calendar is about 2.35 right now with the stock 68.80. That’s not a cheap trade (in premium terms), but the issue is that it’s nearly at the money and so there’s risk of the stock going above that 70 strike too early, before October expires. There are no 72.50 call strikes, which could possible be more ideal and provide some room to the upside over the next few weeks. On the put side, the Oct/Nov 65 put calendar is a nice defined risk way to look for a pullback from these highs. That’s about 1.95 here, it’s the most that can be lost and has some room if the shares pull back in the next few weeks at which point it would be a nice way to be there for earnings as a bearish play.
The options are currently implying about an 8 dollar move from now until November expiration. With the way the stock has gone from $25 to $70 in a nearly straight line in 2016, that may be cheap.