Hosts: Mr. Z (@168MrZ), Victor (@vcmktasa)
Guests: 3X Long Labubu (@labubu_trader), AI Industry Excavator (@QihongF44102)
Crowded Trades, Deleveraging, and Supply Bottlenecks, Semiconductors Remain the Only AI Play in the Public Market
In early June 2026, the AI market, which had risen for two consecutive months, suddenly crashed, with SPX and NDX testing key moving averages. Multiple variables detonated simultaneously: Federal Reserve Chair Kevin Warsh's FOMC debut, concerns about interest rate hikes driven by rising oil prices in the Middle East, the much-anticipated Micron earnings report, coupled with worries regarding the SpaceX IPO's liquidity impact; furthermore, a report from research firm SemiAnalysis questioning the timeline for CPO mass production directly impacted the optical communication sector. Meanwhile, storage and MLCC manufacturers successively raised prices, leading to a dramatic market environment that witnessed both multi-kills and a super cycle.
In this edition of 168X, we invited two highly recognizable investors in AI US stocks and semiconductors on Twitter in both English and Chinese. One is 3X Long Labubu (@labubu_trader), an AI engineer from Silicon Valley, whose expertise lies in market structure, leverage, and options fund flows; the other is AI Industry Excavator (@QihongF44102), a senior AI professional who has progressed from computer vision and search recommendation to the development of Agentic systems, proficient in identifying the underlying bottlenecks being "exploded" using first principles and supply-demand cycles. Both share a striking consensus: Before CapEx stops or before OpenAI and Anthropic go public, semiconductors are virtually the only AI play in the public market. The question is not whether to bottom fish, but when to bottom fish.
I. Silicon Valley Engineers vs. AI Practitioners: One Watch Market and Leverage, One Finds Bottlenecks with First Principles
Mr. Z: First, could the two guests briefly introduce themselves? Labubu, how did you transition from trading and technical analysis into the AI and semiconductor markets?
Labubu: Hello everyone. My primary job is as an engineer in Silicon Valley, and I'm involved in an industry related to AI. I've done some AI-related work, such as inference and training, but I haven't worked in semiconductors myself. So, when I first entered the market, I mainly approached it from a fundamental perspective: as a practitioner, I often take my spare time to study these matters, like what kind of things I am currently using with GPUs and ASICs—what exactly are they? Coupled with interactions with many friends in the industry chain, like excavators and various friends from Asia (Taiwan, Hong Kong, Mainland), I learn about some industry information.
After a few years, I found one thing: even if something has a solid fundamental basis, it can still fluctuate based on the financial and technical aspects. Our advantage as retail investors mainly lies in flexibility; we can take advantage of these fluctuations to profit from short-term trades. So I began as a fundamental investor and slowly learned how to engage in short-term trading. Now I tend to combine fundamental and technical approaches.
Mr. Z: What about Excavator? You also explore supply chain opportunities like MLCC and storage from the perspectives of AI technology and industry.
Excavator: I am also an AI industry practitioner, similar in professional background to Labubu, but he tends to be more focused on the underlying level, while I am more focused on the application level. I studied computer vision in school and have been working on search recommendation systems since. After this wave of Agentic AI emerged, I quickly shifted to this area; my current work involves designing Agentic systems, i.e., the intelligent agents commonly referred to as Codex and Claude Code.
How did I uncover this sector? It relates very closely to my work. Everyone initially focused on training, which naturally inclined you to pay attention to NVIDIA, optical modules, and CPUs, as they impose the most significant constraints on GPUs and interconnects. Gradually, as DeepSeek emerged and inference gained traction, the attention shifted from training to inference—you could see products like Cursor, Claude Code, and Codex, particularly AI Agents rising rapidly. I can see a number: the volume in inference has been rapidly expanding. Rapid expansion brings significant changes: The greatest change is the demand for CPUs and memory, which began around last August and September. This is a big reason why I have been heavily invested in storage and NAND Flash since then.
Recently, there has been a shift: almost everything in the supply chain is rising. You'll gradually find that even analog circuits are starting to rise, and power semiconductors related to power supplies have also surged, and then people began reporting that MLCCs and passive components (capacitors, inductors, resistors) were all increasing in price. The market for passive components is actually relatively small; if you use Codex or Claude Code to search, you can see that the leading players are just a few companies. Comparing this cycle with the one from 2017-2018, you can easily see that this cycle is evidently larger than the previous one; it's just a matter of timing.
This is also a key point I want to emphasize today: When everything is rising in price, CSPs (Cloud Service Providers) are increasing their annual CapEx figures, but due to price hikes, their "real" CapEx may have already been declining, which presents a risk. So how do we respond? One approach is to be a little cautious regarding semiconductor-related stocks at this time. Additionally, we must consider that when everything is rising, other links in the supply chain will feel pressure; CSPs will feel cash flow pressure, and the US and Chinese governments may at this time require everyone to expand production, because only by expanding production can upstream prices come down, allowing CSPs to split one amount into two uses. So if there are adjustments in semiconductors, I believe that after such adjustments, the key focus will shift to semiconductor equipment and materials.
II. How to View June's Pullback: Bull Trap, Early Exit, and "False Risk vs. Real Risk"
Victor: First, I would like to ask Labubu. I have been following your tweets for a long time. You mentioned around Monday that if SPX fell below 7,400, it would guide it to continue declining to 7,200 to 7,260. Indeed, on Tuesday, the market recorded an opening spike that enticed more buyers, then dropped to the low of 7,237 before rebounding towards the end. Yesterday, the CPI met expectations; what is your current view on the market?
Labubu: My view is as follows. In fact, I began mentioning around the end of May, around May 23, that the next two weeks would likely see a significant Bull Trap, ensnaring the bulls and then resulting in a notable pullback. But I did not expect it to happen so quickly.
Let me first explain why I had that judgment. June has many events: Geopolitically, the Strait of Hormuz is not yet fully open, while many refineries in Europe and Asia have already seen their inventories hit rock bottom; the crude oil supply from this strait accounts for approximately 10-20% of the global total (I can't recall the exact number), and if it remains closed when inventories hit rock bottom in June, it could lead to a surge in oil prices. Macro-economically, this oil price shock could drive up CPI inflation and spark fears of inflation; if inflation continues to surge, the Federal Reserve will interrupt its interest rate cuts, and traders are already starting to price in rate hikes before the end of the year. These factors exert tremendous pressure on the stock market as a risk asset.
Additionally, Kevin Warsh's debut FOMC next week adds to this. Since he has appeared infrequently before, there are many speculations regarding him: he could potentially be very hawkish, solely concerned about inflation, and focused on balance sheet reductions, with rumors that he differs from his predecessor in not communicating with the market and being less transparent. The market is particularly sensitive to the "debut" of a Federal Reserve chair: during the last debut (in 2018), some chair’s appearance caused significant losses for the market, hence the heightened vigilance regarding Warsh this time. Furthermore, the potential SpaceX IPO poses questions about its impact on liquidity. Various risks are present.
Based on our past experience, whenever there is uncertainty, the market tends to react pre-emptively to these events, even if ultimately the event turns out to be a non-event. It may turn out later that SpaceX did not drain liquidity, Warsh isn't as hawkish, and CPI isn't as hot, but the market will preemptively hedge due to this uncertainty. As retail investors, we need to react even faster than institutions. This is why we mentioned at the end of May that a major reduction in positions and lowering leverage might be necessary to hedge against risks.
In fact, we also observed significant downward adjustments starting from last Wednesday and extending to last Friday, continuing into today's market retreat. The CPI data released today is very positive: last month’s core CPI showed a month-over-month increase of 0.4, while this time it is 0.2, indicating to everyone that the oil price impact may be a one-time occurrence and that it might have already peaked. Thus, in my view, this CPI is very positive data. Yet even so, the market remains hesitant; today we witnessed a significant rebound, only to be dragged down later due to Trump's actual actions regarding Iran. This indicates that the bulls are quite hesitant and will likely wait for all matters to become clear, especially the FOMC, before a clearer path is visible.
Nonetheless, I maintain a relatively bullish perspective. Since the lows of March and April, we have experienced eight to nine weeks of strong growth in a steep trend; falling into a pit does not represent much, as we have frequently experienced such dips before: on April 19, 2024, early August 2024, and October 10, 2025, all being similar events, yet still within an overall bullish trend. This volatility might simply be driven by emotion or capital crowding, particularly due to leverage pressure in semiconductors, which can lead to a quick drop. So the question is not whether to bottom fish, but when to bottom fish. Personally, I don't have much leverage at the moment, and I'm considering where a good position might be: for instance, if there is a signal of stop-loss near the 50-day moving average, I may consider an initial buy, then add to my position when it returns to key moving averages or support levels.
III. Moving Averages are Just Consensus: Technical Trading Framework and "Scenario + Probability" Betting Method
Victor: You previously posted a risk checklist for June, listing what you consider "false risks" and "real risks," such as the liquidity extraction from the SpaceX IPO, Warsh potentially employing QT (quantitative tightening), and escalations in conflicts in the Middle East. You also mentioned that the real risk is "if these funds believe in false risks and sell," then the market will still decline.
Labubu: Yes, here we need to clarify: the key point is not whether they believe this is a risk, but rather whether they perceive this as uncertainty. One interesting aspect of the market is that regardless of whether the news is good or bad, it has a way to price in those events; however, if it is an uncertainty that can go either way, that's the situation the market dislikes the most.
Many people, especially those on Wall Street, do they have lower intelligence than me? Definitely not. Do they have less information than I do? Also not the case. But they view it as uncertainty and thus avoid taking that risk. So many times when timing the market, the thought should not be "what do you think?" but rather "what do they think?". Personally, I don’t see these as risks; I’m also one of the early investors in SpaceX, and after talking with them, I find none of these risks compelling; however, the market perceives them as risks.
So it becomes somewhat of a mental split: you are an average person, yet you must assume the role of a fund manager in the market, anticipating how a relatively uninformed person would choose? Generally, that choice aligns with the choice made by the majority. As retail investors, we can only flow with that sentiment. Therefore, during times concentrated with risks, we should appropriately reduce our positions or increase some hedges to protect profits. I believe everyone here has probably made a considerable profit this year, and being able to protect profits during a downturn means they could earn even more later on.
Victor: I'm very curious about how Labubu formed this trading framework? I noticed you often use indicators like the 21-day and 50-day moving averages, and also track options fund flows.
Labubu: Essentially, these moving averages represent a consensus. For instance, the 21-day moving average can be understood as the average holding cost for all traders over the past 21 days. I like to use it not because these indicators are particularly magical, but because everyone else uses it. When everyone employs the same type of indicator, it becomes a consensus. This is why we often see the fascinating phenomenon of prices accurately reversing at the 21-day or 50-day moving averages, or once breaking below the 50-day moving average, they continue to descend without recovery. It is not a golden standard, but simply a market consensus, a statistical measure based on past experiences. For example, over the past ten years, during smooth and linear upward trends, you would find that the major market had never closed below the 21-day moving average.
Therefore, I commonly use elements that are common among everyone, the more common the better, avoiding overly fancy tools; they only assist me in identifying entry and exit points: moving averages, gaps, highs and lows, and key levels. For example, why is the 7,330 level significant? Because back on May 19, we saw significant turnover there, making it a market-recognized fair price. As a trader focused on technical analysis, you need to firmly grasp these points. How did I develop this sensitivity? Through trading futures, as futures are purely point-based; for example, in ES, one point equals $50, and in NQ, one point equals $20, which creates a rich and memorable experience.
I noticed that on Tuesday, the market inexplicably rose to around the high of Monday, specifically the 7,480 level, which seemed unreasonable to me, so I cleared all my leveraged positions, and sure enough, it quickly fell back down, confirming my doubts. This is merely a consensus: if it is a market brimming with optimism, it would easily break through the high and stabilize, climbing higher; but it didn't happen, and instead, at that high point, panic set in, causing many to think, "I finally got out," prompting them to flee.
Victor: On May 29, SPX also saw a large volume spike, rallying for another two days before initiating this recent pullback.
Labubu: Correct. Thus, using ATR multiples together with observing daily price-volume relationships are great methods for profit-taking. As for stop-loss methods, they generally revolve around key support levels, gaps, or moving averages, so they're somewhat simple. In fact, the most challenging part for most people is often the profit-taking, not the stop-loss; "knowing how to sell is what makes you a master." Many have experienced this bullish market, having witnessed stocks that soared, similar to early year stocks such as Samsung, SK Hynix, SanDisk, and even those crazy stocks like ARM. Hence, recognizing volume peaks and ATR deviations are strategies I commonly use to ensure I protect profits while not unnecessarily selling out too soon.
XIV. Advice for the Audience: Living Through the Year of the Fire Horse and the Signals Behind Model Price Cuts
Victor: We have discussed for almost two hours; finally, I want to ask, 2026 is the Year of the Fire Horse in Chinese metaphysics, which suggests a year of great changes and volatility. We have also seen this year's market rapidly fluctuating. At this point, and considering the second half of 2026, what do you recommend for investors to follow the trend and avoid harm?
Labubu: I feel that starting from the movement over the past couple of days, the second half of the year might present a very choppy trajectory, with fluctuations such as several days of rises followed by several days of declines, which can be frustrating and discomforting. However, the overall market is in an uptrend. Therefore, until we see clear signs of CapEx halting or decreasing, particularly for those trading semiconductors, I think it’s essential to strengthen one's ability to weather pullbacks during this process. Regarding some comments from SemiAnalysis, some have partial truths but also contain inaccuracies—they can be quite "disgusting"; you would have to question if they have an ulterior motive. There might be interests behind them wanting to obtain assets at lower prices.
Thus, I think during this period, one should reduce leverage usage while taking care of selected stocks and properly executing profit-taking and stop-loss strategies. For sciences, equipment, and optics, even though Excavator has expressed doubts earlier, I still regard the optics segment as a strong sector, and for the MLCC, I believe we can earn significant profits from them. Before CapEx stops or decreases, or before OpenAI and Anthropic go public, semiconductors remain the only AI play in the public market.
Victor: Do you consider the revenue growth rate of model providers a potential risk point? Recently, there has been discussion about whether models have a "dumbing down" issue.
Labubu: I would not think so; rather, I see it as a good thing. The "dumbing down" reflects a lack of computing power, but consider the news today, for example, OpenAI aims to lower prices to compete with Anthropic by reducing token prices. If I can spend $100 on your service and receive a downgraded model, I would, of course, switch to another service; additionally, the migration cost to transfer from Claude Code to OpenAI's Codex isn't that high—so long as you've developed many skills, migration can be done swiftly with AI assistance. Ultimately, this will resemble the competition among cloud service providers in the past, where token unit prices are set to decrease, which is good news for downstream applications as it allows the entire AI narrative to progress and expand significantly.
With increased competition among model vendors, if Anthropic still charges $200 for a relatively weak model, but OpenAI offers a smart model for $100, I will definitely opt for that. Hence, their overall revenue will still exhibit exponential growth, but the unit price of tokens will undoubtedly drop. However, in contradiction, this also implies that the upstream semiconductor’s routine notion of "relying on price hikes" will eventually be rapidly challenged, and the speed at which such price decreases ripple upstream along the supply chain may be quicker than most anticipate. This aligns with what Excavator mentioned.
Victor: Last, a lighter question, why is your ID "3X Long Labubu"? Why do you hold Labubu in such high regard?
Labubu: Because my daughter loves Labubu. In the past when Labubu was relatively hard to find, I spent considerable time gathering them for her, but that's no longer the case. Ultimately, it’s because my daughter loves Labubu; whenever new ones come out, I buy them.
Mr. Z & Victor: Thank you very much to Professor 3X Long Labubu and AI Industry Excavator for joining us today. They have shared a wealth of insights into the current market from technical, data-driven, to fundamental and macro perspectives over the last two hours. Thanks also to every audience member listening along. If you enjoyed this episode, please follow both guests and subscribe to 168X on X and YouTube, sharing this program with more friends interested in AI, semiconductors, and trading. See you in the next episode.
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。
