November Stock Musings: Is This the Beginning of an End, or the End of a Beginning?
2 weeks ago
Hello everyone, and welcome to November.
This week's title says it all: lots of questions, not many answers.
Starting from that Friday at the end of October, the market saw another round of selling. One fairly notable recent disruption came from Ramp. Its previous tracking of Anthropic's ARR was pretty accurate, perhaps because the platform is larger and has more data. This time it noted that the value per token for Anthropic's Opus 5.5 is on the low side and has yet to return to its July peak, and the market subsequently began another round of selling.
The election is now settled. Based on how the market currently sees it, the power shortage mainly affects 2028, with relatively limited impact in 2027, since the permitting and construction schedules for the relevant projects are basically locked in. But semiconductors and related sectors typically react 12–18 months ahead, so there's still a gap between the "distant prospect of a power shortage" and "the reality of a power shortage," and the market will keep trading this expectation back and forth.
In the short term, everyone is once again putting the big hyperscalers on a pedestal—AWS, for example, because it has the largest overall scale and is guiding for strong growth. The problem is it only gave two years of guidance. If 2028 permits can't be obtained, or become harder to obtain, can it still achieve 40% growth? Or, by the time the 2028 election comes around, will this become a very sensitive issue? So there are still many questions, with no clear answers. Personally, I think most of the time people are just cheering on the larger trillion-dollar companies. But the questions still outnumber the answers: the power gap existed before the election, and no matter how you measure it, you can't get around it.
Based on current information, Texas has suspended issuing new related permits since August and is conducting an audit, expected to be completed in December. Abbott has also shifted from being very supportive of AIDC to restricting some project construction under campaign pressure—his stance has changed. So regardless of who wins or loses, it's only a relative narrative change. The power shortage problem still exists, and the market is currently mainly viewing it through the lens of 2028.
What I've observed is that everyone has moved from previously controlling position sizes to actually reducing exposure. This is a painful and drawn-out process. Early October was still fairly optimistic, then came a bloody few days, one bloody week, then a few bloody weeks—will it turn into a few bloody months? Validating the whole narrative will take time.
The core view hasn't changed: after Coding, when the next application scenario will open up, and when the next wave of investor capital will come in, remain the key questions. These questions have persisted ever since this "Pandora's box" was opened at the end of last year. Institutions currently have plenty of room to be underweight, at least in the short to medium term.
Why does everyone think Anthropic's IPO is so important? Because it has already raised several rounds, and VCs have come in round after round. Typically, VCs need to first distribute returns to LPs through a project's IPO and subsequent exit before it becomes easier to raise the next fund.
And what are these VCs investing in now? Many are new labs—also the highest-risk batch—betting on whether the models can be built and whether SSA can be built.
The market's fairly consensus view right now is: if these new labs can't raise their next round, or if fundraising becomes difficult, GPU rental hours will come under pressure, and related CapEx may also be cut. This is a transmission chain everyone is watching very closely.
But this process takes time: IPO, exit, raising a new fund, then investing in new labs—none of these steps happen overnight. On the road to AGI or the next application scenario, it's quite likely that news will keep emerging along the way about companies' fundraising being blocked, CapEx orders being cut, capacity delays, and data centers scaling back or permit freezes.
So during this period, you still need to watch your position sizing. Long-term imagination is one thing; whether the funding can keep coming and whether projects can keep moving forward are the issues that need continuous tracking right now.
Texas is pausing new permits in August and won't complete its audit until December—that timeline needs close watching. If the December audit comes back tight, 2027 project schedules may need to be re-evaluated as well. The market is currently only pricing in the power shortfall for 2028, and this gap in between is easy to overlook, so it's still wise to leave some room in position management.