My Chip 'Heavier' Plan as Nvidia Scrambles to Meet Surging H200 China Orders
Rumors of the death of the AI-focused trade based on capex spending have been greatly exaggerated. Here's my semiconductor strategy for 2026.
You're reading 0 of 1 free page.
Register to read more or Unlock Pro — 50% Off Ends Soon
Now, that's news!
Early on Wednesday morning, Reuters reported that Nvidia (NVDA) was "scrambling" to meet Chinese-based demand for its H200 AI GPUs. Nvidia is said to have approached its main chip foundry, Taiwan Semiconductor (TSM) , to see if production levels could be increased.
The report states that more than 2 million H200's have already been ordered by mainland Chinese clients for calendar year 2026. It also states that Nvidia only has something like 700,000 of these chips in inventory.
Reuters also reports that a tertiary source claims that work on the increased production of said chips will begin sometime in the second quarter of 2026. If one goes just a little further back in time, say to Tuesday night, the South China Morning Post reported that TikTok parent ByteDance, which is obviously a Chinese client of Nvidia's, had made plans to spend roughly $14 billion on AI-capable chips over the coming year and planned to buy those chips from Nvidia.
What's This All Mean?
Well, I think it's quite clear that the rumors of the death of the AI-focused trade based on capex spending have been greatly exaggerated. I mentioned in Wednesday morning's Market Recon that I would welcome a broadening of the 2025 equity market rally, but not at the expense of big tech or this AI trade. Sharing the wealth would be great, but tech still has to be if not the leader, a leader.
As for the Designers...
Struggling to keep up with demand is obviously great news for Nvidia. That said, frustrated Chinese customers may turn to other sources, both Chinese domestic and imported chips to satisfy their needs. That's a huge positive for a firm like Advanced Micro Devices (AMD) as well, as that firm is Nvidia's closest, if not close competitor in the space.
Should this condition impact global markets as well as a focus is placed on meeting Chinese demand, a firm like Broadcom (AVGO) could also see improved fortunes.
As for Memory....
So, Micron Technology (MU) , Western Digital (WDC) , Sandisk (SNDK) and Seagate Technology (STX) all had great years in 2025 due to the growing need for memory and data storage due to the AI revolution.
I played this group poorly this year. I did benefit nicely for a good part of the year in Micron, and had one nice trade in Western Digital (WDC) , but for the most part, I could have done better. The memory/storage trade is not over.
As for Equipment Providers...
I was long Lam Research (LRCX) for much of the year and did benefit from that trade. KLA Corp (KLAC) and Applied Materials (AMAT) are the other key players in the space. Both are good stocks. I probably should not have sold my LRCX, as I did my MU, but I have become concerned that my portfolios had become too chip heavy and too heavy.
As for Manufacturing...
While as the world's leading and most skilled foundry, Taiwan Semiconductor is in the catbird's seat, this does open the door for Intel (INTC) , which is a name that I was working my way out of, to get its head out of its tail and play catch up.
Until Intel is seen as a viable competitor in this space for the manufacture of the most complex and smallest chips, this remains Taiwan Semiconductor's game. In addition, should TSM soon move some of this high-end production to its facilities in Arizona, that diminishes Intel's leverage as a domestic U.S. foundry.
My Plan Going Forward
I need more chips. No, not chips with salsa. I have been long both AMD (+78.5%) and NVDA (+36.6%) all year. That's not about to change. Both Jensen Huang and Lisa Su are set to speak from CES 2026 in Las Vegas on Monday. I got into AVGO late in the year. I'm not leaving. I'm still in INTC small, but the position means nothing to me. I need to be in TSM.
Pushing further out, I need to be back in the equipment providers, and back in memory. I prefer LRCX but would settle for KLAC. I prefer MU but would settle for WDC. The problem is that there has not been a real dip in these stocks since I got the bright idea to sell them.
Bottom Line
Yes, I am chip heavy. I am pretty sure that I need to be chip heavier.
Have a great new year, everyone. Sarge out.
At the time of publication, Guilfoyle was long NVDA, AMD, AVGO and INTC equity.
