Weekly|Less Pessimistic than Citrini, NVDA, LLMs, Optics, Software and Space stocks earnings
It’s probably not a proper Weekly to not start with some color on Citrini’s piece that sank everything the past Monday. Kudos to Citrini for his piece, which was very well written. The issues discussed are ones we have frequently discussed in our LLM Series since last year, though we never presented such a grand narrative. However, as we have already laid out in X earlier, we are not as pessimistic mainly for two reasons:
Productivity improvements actually cause the demand reduction mentioned by Citrini. Demand reduction driven by productivity improvements is a special economic scenario, and most of it can be offset by macro policies, unlike other demand-reduction scenarios.
After working hours decrease, humans will have a significant amount of time for consumption and entertainment. You cannot assume that AI will replace all friction and consumption; humans would become vegetables. Imagine the lifestyle patterns of people who now have plenty of free time: they spend more time on short videos, games, and shopping. In reality, we need to refer to the economic systems of Norway and Saudi Arabia, which may be highly similar in the future. When profits increasingly concentrate at mega companies, systems similar to Norway’s and Saudi Arabia’s oil systems or sovereign wealth fund systems will emerge. More transfer payments will be distributed to people in the form of high welfare benefits. Human working hours will continue to decrease, but consumption time will continue to increase.
Also, there are numerous discussions on policy lags. We actually don’t expect a “gray rhino” scenario.
Just as AI movies like The Matrix have existed for 40 years, we’ve only really started to touch AGI in the past three years. As for what kinds of economic and technological impacts AI will bring, humanity has been discussing them in all kinds of literary works for decades. This doesn’t require any additional background framing or “awareness building.” Society doesn’t really have a directional divide in understanding; the bigger issue is uncertainty about the speed and the eventual magnitude of the impact. There isn’t even much of an attitudinal divide—everyone knows this is the most important issue, and so do different actors across society. This doesn’t look like the typical gray rhino setup.
We are not trying to argue “this time is different.” We only want to emphasize that AI will be the most important trend for humanity over the next 10 years, and even the next 100 years, so people’s focus will be completely different from that during a financial crisis, a debt crisis, or a localized war.
Generally speaking, trying to prove ‘this time is different’ is a foolish approach. It’s a weak perspective, you’re always hoping things aren’t as bad, or worrying that the next narrative will upend your own views. This is always the case when investing in stocks. But in today’s era of dramatically improved productivity, solutions outnumber problems; isn’t your speed of thinking through problems now also 10 times faster than before? Of course, we don’t rule out that ultimately an AI Tax might be adopted as a viable method of redistribution, and if that happens, it would be understandable.
This week’s reports
NVDA review - Definitely disappointing price action after a strong print. Feedback on the poor reaction we heard include persistent GPM pressure and an uncertain competition landscape in inferencing, among others. We think in general, NVDA share price is more about how much incremental capital is available out there to rotate into this $5tn giant, which is heavily reliant on overall market risk appetite. GTC is the next catalyst to watch and we will have our preview next week.
Update to US/China LLM competition - Since we published our previous article comparing Chinese and U.S. foundation models, many readers have written in asking how to interpret the recent wave of Chinese open-source model developments, and whether a second “DeepSeek moment” could emerge. Therefore, we conducted a follow-up deep dive into this question.
The Defining Trends from ISSCC 2026 - This is a reproduction of a great technical piece by our friend at Silicon Task, which is closely aligned with our line of research. We will be releasing a few more deep dives on optics names soon.
The hottest topic within optics is clearly CPO. Recently, the market has shown heightened attention toward discussions between Broadcom and CSP customers regarding CPO-related solutions. A brief clarification is warranted here:
At present, the industry’s definition of “CPO” remains relatively broad and somewhat ambiguous. The solution being advanced by Broadcom, in collaboration with cloud service providers, is, in essence, an NPO-type open-architecture optical interconnect approach rather than the traditional CPO-led, tightly controlled approach of switch chip vendors.
In many discussions, NPO is often categorized under the broader CPO umbrella. The underlying objective in both cases is similar: to shorten electrical interconnect distances and mitigate high-speed signal loss across PCB traces.
Overall, Broadcom is primarily promoting an open architecture with NPO rather than a highly integrated, vertically closed CPO model. This strategic direction is, in fact, constructive for ecosystem participants that possess NPO design, advanced packaging, and supporting integration capabilities.
MDB preview - The most interesting development in the past quarter is that LLMs are moving away from their previous PostgreSQL-only approach by adding NoSQL to their database architecture. This is the biggest validation of MDB’s AI beneficiary status to date, but we are mindful of its relatively high valuation compared to other software.
SNOW and CRM reviews - two different sets of results, but similar price action. SNOW guided growth acceleration over the next 3 quarters in 2026 with strong margin guidance as well, while CRM’s results are still underwhelming. We would look for divergence in stock performance after the beta unwind settles.
Space stock results - RKLB and RDW. For RKLB, while the delayed Neutron rocket launch remains the core variable holding the stock back, we like its timely strategic acquisition of OSI and Mynaric, and remain positive on its long term outlook. On the other hand, market’s patience on RDW could be more limited as it is not able to boost profitability with revenue expansion.
