Investment Notes: In-depth Review and Analysis of Lucrative Targets from the 2026 GTC Conference!

After watching the entire Nvidia 2026 GTC conference, it was quite shocking. Mr. Huang is repeatedly instilling a new understanding that in the future, the competition in #AI will no longer be about big models, but rather about Token factories. Especially after the explosive popularity of Openclaw🦞, people are gradually recognizing that the enormous revenue generated by Token consumption will become the core pillar of AI companies in the future.

Currently, there is just a small crayfish🦞, but in the future, there could be little pandas🐼, little hamsters🐹, and so on, resulting in hundreds of thousands of similar Agents. The era of inference explosion has arrived, and it is expected that in the next 3-5 years, Token consumption will achieve a growth factor of 1000 to 10000 times, while the continuous demand for computing services will lead to sustainable Token consumption.

If we say that software companies are coming to an end because of AI Agents, and the previous SaaS subscription model has become history, then in the AI era, the AaaS model will become the future, where collecting Token fees will bring massive profits for businesses. The barriers to entry will also be higher; once a certain model is chosen, the continuous feeding and consuming of Tokens are essentially a unique accumulation of data and self-growth. Your exclusive Agent at this moment grows like a baby, and it will be difficult to topple it down and raise it anew unless absolutely necessary. In the end, what we are competing on is the GPU computing power of the giants, which is also a contest of electricity and data centers, with every data center in the future being a unit that generates fees.

If you understand this new logic instilled by Mr. Huang, then the more than 600 billion capital expenditure budgets of the giants this year will make sense; after all, the more you invest now, every data center being built will turn into an endless Token printing machine in the future, competing for GPUs, electricity, construction periods, and memory chips—everything seems to be justified!

So next, let's talk about the favorable US stock targets👇

1️⃣ "Copper and Light Coexist": The Golden Decade of Connectors

At this conference, Jensen Huang provided a definitive tone: "Use copper inside the cabinet and fiber outside the cabinet." Previously, there were rumors in the market that Nvidia's new architecture would lead to “copper going in and light going out”, causing significant concern in the market. This firm stance basically cleared the gloom over the optical module sector.

Currently, in terms of development direction, to meet physical limits of cost and energy efficiency, copper cables inside the cabinet will still be the focus; but with the full production of the Spectrum X CPO (co-packaged optics) switch, the breakthrough point for optical communication has shifted from “modules” to “integration”, which can be directly integrated into ASICs, overcoming the limitations of electronic signals in large-scale AI data centers and achieving higher efficiency in transmission. The collaboration between Nvidia and TSMC on CPO will continuously benefit these optical signal detection and laser component companies.

Favorable targets:

🟡#AAOI (Applied Optoelectronics), this company assembles optical chips and devices into "usable optical modules" and sells them to data centers.

It's analogous to 👉 Foxconn, which integrates components into a sellable iPhone.

🔵 #LITE (Lumentum), this company mainly produces core devices that "make light move" (lasers, modulators).

It's analogous to 👉 the A-series chips in the Apple iPhone, which are key components that determine performance limits.

🟢#AXTI (AXT Inc), this company provides the underlying materials (InP, GaAs substrates) necessary for manufacturing optical chips.

It's analogous to 👉 selling rare earths or silicon wafers; while the narrative isn't glamorous, all high-tech (optical modules) cannot do without them.

🟣 #TSEM (Tower Semiconductor), this company is a specialized foundry that helps others manufacture photonic/analog/RF chips.

It's analogous to 👉 TSMC, which produces various nano-model chips (but specializes in niche high-tech processes).

2️⃣ Rubin Architecture and "Year of Storage": HBM4 is the Game Changer

The Rubin architecture is not just an upgrade in process technology; its biggest change is the introduction of HBM4 and 100% liquid cooling. This time, Jensen Huang mentioned that the capacity of HBM4 will determine the rate at which Rubin production lines expand, effectively pushing storage chips from the "cyclical stock" logic to the "fundamental AI consumables" logic. No wonder Mr. Huang has been in South Korea recently since Samsung and SK Hynix are absolute leaders in memory chips.

At the same time, removing cables and implementing 100% liquid cooling means that the infrastructure of data centers must undergo significant reform. Companies related to liquid cooling will directly benefit from this; we have emphasized several times in previous tweets, and since some stock recommendations have already increased by over 50%.

Favorable targets:

In the storage field:

🟢 #MU (Micron Technology), this company we have recommended multiple times around $150, is an all-round player in DRAM + NAND, and AI memory HBM is currently catching up.

It's analogous to 👉 "The American version of Samsung", but smaller in size and slightly slower in rhythm.

🟡 #SNDK (SanDisk), this company we have also recommended multiple times, focuses on NAND flash (SSD, memory cards), leaning towards consumer and storage devices.

It's analogous to 👉 "The Western Digital of the storage industry", specializing in hard drives and SSDs, generally not dealing in high-end DRAM battles.

🔵 Samsung (Samsung Electronics), the undisputed global memory giant, covering DRAM + NAND + HBM across the board, excelling in both technology and production capacity.

It's analogous to 👉 "The TSMC + Apple fusion of the memory industry", capable of achieving the strongest technology while also facilitating large-scale deliveries.

🟣 SK Hynix, currently the absolute king of HBM (AI high bandwidth memory), has captured the most lucrative segment of AI.

It's analogous to 👉 "The core ace in Nvidia's supply chain during the AI era", specializing in feeding AI GPUs.

Since US stock accounts cannot purchase Korean stocks, I recommend this US stock ETF, the Korea ETF (code: #EWY), which has 22.46% of its holdings in Samsung, 19.39% in SK Hynix, and other holdings include many excellent Korean companies, such as Hyundai.

Liquid cooling:

🟤#VRT (Vertiv Holdings), this company provides a complete solution for data centers, offering "power + cooling + infrastructure", especially for AI data centers.

It's analogous to 👉 "The air conditioning + power system general contractor for data centers", which, although not engaged in computing power, determines whether computing power can operate normally.

3️⃣ OpenClaw and Token Compensation: The "Windows" of the AI Era

This was the point that amazed me the most. Jensen Huang positions OpenClaw as the operating system for Agent computers, a positioning that transcends the logical boundaries of systems and software in the existing internet era. It means that in the future, we may no longer have the concept of software apps, and we need to gradually familiarize ourselves with hiring individual Agents.

"How many Tokens are included in your Offer?" This question reveals the essence that computing power equals wealth in the future. As the cost of computing power decreases, with the efficiency gains from Rubin + Groq 3 LPX, the proliferation of AI agents will lead to a new surge in AI cloud services.

Favorable targets:

🟢#IREN (Iris Energy), this company was originally a BTC mining company and later upgraded its mining sites into an AI computing power leasing platform (GPU cloud) thanks to its own low-cost electricity + data centers.

It's analogous to 👉 "Transforming a Bitcoin mine into an AWS computing power rental factory."

🟡 #CIFR (Core Scientific), this company just crawled back from the brink of bankruptcy and transformed into a "managed + AI computing power" mining operator.

It's analogous to 👉 "Shifting from a coal mine owner to a data center landlord, while also managing servers for others."

🔵 #NBIS (Nebius Group), this company offers a pure AI cloud platform (GPU cloud + AI services), not stemming from a mining background but more technology-driven.

It's analogous to 👉 "The AI version of AWS or CoreWeave (more tech-driven)."

4️⃣ Spatial Computing and Feynman Architecture: Aiming for the Stars

The next-generation Feynman architecture's 3D stacking and "space chip" plan indicate that Nvidia’s vision is no longer limited to Earth. This may not just be a gimmick, as every time Mr. Huang makes bold claims, he delivers. This space chip initiative appears to be the extreme extension of edge computing. Running AI in high radiation and extreme space environments imposes extremely high reliability requirements on chips and satellite links.

Favorable targets:

🟢 #RKLB (Rocket Lab USA), this company provides an integrated solution for "small rocket launches + satellite manufacturing + space services."

It's analogous to 👉 "The space version of SF Express + Foxconn", capable of both delivery (launching) and satellite manufacturing.

🟣 #ASTS (AST SpaceMobile), this company uses satellites to directly provide 4G/5G signals to mobile phones without the need for ground-based stations.

It's analogous to 👉 "The space version of China Mobile", directly giving you signals from the sky.

The above 👆 outlines the favorable targets we summarized from the GTC conference. If you have any suggestions or additions, feel free to share in the comments for discussion! 🧐

Currently, the companies mentioned above can basically be found on #MSX. To trade US stocks, I choose the #RWA US stock tokenization platform #MSX to jointly invest and participate in the US stock market: http://msx.com/?code=Vu2v44

Early US stock investment fans and partners can private message me. After filling out the form, you can enter the US stock discussion and exploration community for free (limited to 10 people per week, assistant review may take a little time, thank you 🙏)!

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。