The Bull Case for Broadcom Amid a Fickle AI Trade
The Current Situation
Broadcom's stock reached an all-time high of $495 on June 3, but it has fallen sharply since then. Shares are down roughly 25% from that peak, trading around $366 and change, off another 3.4% on the day in question. This raises the obvious question: with the stock off more than 20% from a peak reached not long ago, what exactly is the story?
The most important point to establish is that the decline has very little to do with how the underlying business is performing. The business is doing phenomenally well. Broadcom is growing a very large business by 50%, and is expected to continue doing so for the foreseeable future. The weakness in the stock is therefore not a story about deteriorating fundamentals — it is a story about market psychology and the rotating nature of the AI trade.
Why the AI Trade Moved On
The AI trade is fickle. It started with Nvidia, then expanded to include Broadcom, then added AMD, and eventually swept in the entire semiconductor complex and beyond. Having broadened that far, it has now simply moved on — away from Nvidia and away from Broadcom.
The most recent trigger was a matter of comparison. Nvidia, AMD, Intel, and Micron all reported phenomenally strong quarters. Broadcom reported a strong quarter too — but a strong quarter looks less impressive when set against a string of phenomenal ones. That relative shortfall, rather than any actual disappointment, is why the AI trade rotated away from Broadcom. The move is about the fickleness of the trade, not the health of the company.
A related dynamic is visible day to day: the market's attention has moved beyond the hyperscalers and the big spenders toward more of the "bottleneck" parts of the AI supply chain. On any given session there is a kind of ping-pong match between the hardware names, the software names, and the "Magnificent Seven," with the groups falling in and out of sync.
When Will the Groups Sync Up Again?
A natural question is when this rebalancing will occur — when the hardware, software, and mega-cap groups will start to move together again. The honest answer is that this is hard to predict. But there is a knowable mechanism behind it.
The large spenders — Microsoft, Amazon, Google, and Meta — are telling us directly that they are getting returns on their investment. The market currently seems to believe something else. But these companies say they are selling data center capacity at a very large markup relative to what it cost them to build the data centers. They are signing three- and five-year deals that lock in those returns, and they intend to keep doing so for as long as AI demand persists. As their growth continues to accelerate, and provided they can sustain it at stable margins, the market should eventually return to these big spenders. That outcome, however, depends on continued progress in AI.
The Evidence That AI Demand Is Real
Amid constant news flow, the single most important recent data point is the scale of actual AI spending. It now appears that Anthropic has crossed a $60 billion-a-year run rate, and OpenAI has crossed a $40 billion-a-year run rate. Those two companies combined therefore represent about $100 billion in real, actual spending on AI tools — a figure that was zero just three years ago.
This is concrete proof that the technology is genuinely being used. People want to incorporate AI into their personal lives and are using it at work, and companies are buying more and more AI capacity. That growth will not move up in a straight line, but it clearly demonstrates real utility and a real willingness to pay. Crucially, those buyers are paying Amazon, Microsoft, and Google. As this revenue becomes more visible, as it accelerates, and as the returns materialize, investors are likely to shift back toward those stocks.
Jalapeno, OpenAI, and the IPO Question
How much of Broadcom's specific move is tied to the "Jalapeno" announcement made this week with OpenAI, especially given New York Times reporting that OpenAI could potentially be delaying its IPO?
Both OpenAI and Anthropic face a genuinely complex IPO decision. On one hand, capital is available now, and they could put that capital to work growing their businesses. On the other hand, remaining private carries real advantages: less pressure to turn a profit, no obligation to report numbers publicly, and the freedom to focus much more on the long term. So they are weighing whether to go public at all.
Layered on top is significant game theory. OpenAI does not want Anthropic to go public first and then find itself having opted not to — a scenario in which Anthropic is public and OpenAI is not. Both companies also just watched SpaceX come down sharply from its first day of trading, which likely chilled any intention of an IPO this year.
Importantly, delaying an IPO does not mean delaying capital spending. The relevant players just raised $122 billion, so they have the money to spend on compute. And part of that compute spending plan is making their own chips.
This is where the bull case sharpens. Building their own chips is, strictly speaking, a little superfluous — they do not really need to do it. But they are going down that path anyway because, long term, they believe owning custom silicon will give them an advantage, or at least put them on par with Google, which makes its own chips. OpenAI and Anthropic both want to do the same — and they are probably both going to do it with Broadcom.
The Outlook Over the Next Several Quarters
The whole landscape has become a story about vertical integration, which is why Wall Street has effectively split the Magnificent Seven into two separate buckets.
For Broadcom specifically, the big question over the coming quarters remains the cycle. If AI's current spending cycle continues, then all of the large companies — Microsoft, Amazon, Google, and Meta, as well as OpenAI and Anthropic — will want to diversify away from Nvidia. The main way to do that is with Broadcom. So as long as the AI cycle continues, Broadcom will keep growing at these rates, and the AI trade will eventually come back to the name as well.
A Sample Options Trade on Broadcom
From a technical standpoint, Broadcom is lagging behind some of the other names in the chip sector. The sector itself remains strong — quite an outperformer, doing better than the S&P 500 as a whole — but Broadcom is lagging not just the SMH semiconductor index but the broader tech sector too. That relative weakness argues for expectations that are positive but somewhat muted and tempered.
On that basis, a representative trade is a short put vertical out to the July monthly expiration: a minus-one July 17th 335/330 put vertical, taken in at a $1 credit (the price available at the time the trade was structured). The setup is 21 days out with a neutral-to-bullish outlook — a softer target than outright buying a call. The goal is not to beat a price target but simply to stay above the break-even and keep the credit collected.
The payoff profile: maximum profit is the $1 credit received, and maximum loss is $400 if the stock falls below the 330 protective long put. That works out to roughly a 1-to-4 risk/reward, a fairly common structure traders use for this kind of play. The probabilities favor this trade because the 330–335 zone sits about at the edge of the options market's expected move range and is also an old area of support. The break-even is around 334, about 9.7% to the downside — roughly in line with the expected move — so the trade is essentially a bet on the probabilities playing out at that supportive area.
