Shares of Broadcom (AVGO) tried to recover after a drop on Wednesday. The stock fell more than 4% after Marvell (MRVL) expanded its work with Google (GOOGL) on custom AI chips. This raised worry that Google is spreading its custom chip work away from Broadcom for its TPU business.
Broadcom is not being pushed out
The Marvell-Google news favored Marvell, and Marvell rose while Broadcom fell. But Broadcom keeps its own strong position. Its Google partnership runs through 2031. It also works with Meta (META) on custom AI chips, and it still has a deal with OpenAI. The demand is not going away.
According to Market Watch, Bernstein says the bigger story is the sheer size of AI demand. There is enough business for everyone in a market that stays compute constrained (short on computing power). Shares of Broadcom are up about 25% year over year, and some investors still hold the bull case. This is not a zero sum game.
Firms like Alphabet (GOOGL) are likely trying to spread out their suppliers so they do not get caught short of product. Behind the scenes there is a hard search for enough product. That is what the market missed on Wednesday. Buying Marvell on the news makes sense, but selling Broadcom off that hard does not.
Timing and earnings
Broadcom reports earnings September 2nd, late in the overall cycle. The key question: has anyone anywhere shown any drop off or slowdown in demand? Nvidia (NVDA) reports in about 6 days, the first big test, then Broadcom on September 2nd. The market is reacting to a story, and it does not change the big picture.
Looking at the charts, semiconductor and memory stocks put in a near-term bottom in the second or third week of July and started to recover. Broadcom sits today at the exact same level as that July low, around 363, even though many of its competitors outperformed it. The stock rallied for a bit, then pulled back. It sits at an inflection point. It will be worth watching if it picks up support into earnings, with Nvidia's report playing a part.
Analysts expect earnings per share of about 3.24 on revenue of about 29.5 billion dollars, an 85% year-over-year jump on that revenue forecast. How much is already priced in remains to be seen. On valuation, Broadcom trades at about 30 times forward price to earnings. Many other stocks trade at similar levels without the same growth potential. Its PEG ratio (price to earnings divided by the estimated growth for the next 3 to 5 years) is around one, which points to the more undervalued range. Value investors may start to look at the stock into earnings. Broadcom has probably been a little overpunished, and semiconductors as a whole have had a rough week or two.
The bear-leaning trade (neutral to bearish)
This approach sits in a passive to bearish frame, looking above the market and selling. Because earnings land September 2nd, the trade uses the August 31st expiration, which expires before earnings. Steps: look at the expected move, then sell the 385/395 call vertical, collecting about 1.40 (now trading maybe a dime lower). This has an 80% chance of finishing out of the money.
It works as a short-term hedge on a long position. The stock can still rally and the trade can still profit. It is a high probability short call vertical that collects theta (time decay) and is risk defined. The idea is not to overcommit, giving the stock room to recover ahead of Nvidia's earnings and Broadcom's own September 2nd report. When selling premium into an event but keeping the trade outside that event, the key is asking what catalyst could move the stock enough to put the vertical at risk. Right now there is not much strength in semiconductors, and there is no clear catalyst before earnings.
The bull-leaning trade (broken wing butterfly)
This trade goes into the earnings cycle and is a make or break bet. It plays two things. First, possible support forming around the 360 level, which the stock has bounced off twice, sitting right at the 200 day moving average. Second, the call skew: investors are paying up again for upside calls compared with the last 30 to 90 days, so many people are betting on an upside move.
The structure is a broken wing butterfly on the September 4th expiration: buy the 360 strike, sell two of the 390 strikes, and buy the 410 strike. That is a 30 dollar wide long vertical, with cost reduced by selling a 20 dollar wide short vertical. It expands to its highest value at 390, close to and a bit inside the expected move for September 4th. Cost is about 8 dollars (7.90 debit), so it is not cheap, but it offers almost a triple if the stock reaches 390.
Because of the call skew, the two 390 calls can be sold for a higher price than normal, taking advantage of the bid to the upside. If the move happens and the stock gets close to 390-395, two things help the trade: the in-the-money 360 calls gain delta as the stock rises, and implied volatility drains out of the 390 strikes, which helps because the position is short two of them. The 410 leg is bought to keep the trade risk defined. Done as a 1x2 ratio it would carry open risk, but the 410 caps the loss if the stock rips higher.
This butterfly does not hurt you if the stock goes through the strike and past it. You will not peak out on earnings, but you stay profitable. That fixes the biggest problem with a butterfly, where a move blowing through the strike wipes out all profit. This one keeps a fair amount of it.
