Your Catchy English Title
2025: NarrativeA year of being schooled by the market
"Let's talk about the AI bubble, chips, and black swans from a Wall Street perspective—how we missed Google and went looking for those hidden marginal opportunities in the midst of the frenzy."
Looking back at 2025, what struck me most was the intensity of that "narrative shift." At the beginning of the month, everyone was fixated on fundamentals, thinking Google's lunch would be completely eaten by AI. By mid-year, it was the companies considered laggards that showed surprising resilience.
I have to admit, our judgment on Google was far too pessimistic. Wall Street once thought social and search were the easiest for AI to disrupt, but after July, Google's performance was unexpectedly strong, while some so-called "Junk Stocks" performed even better. Behind this is a brutal reality: as the likelihood of Nvidia doubling again diminished, capital began frantically searching for marginal changes outside of compute clusters.
Marginal Beats Core
Capital is no longer just staring at GPUs; it has shifted to the power narrative, the memory track (HBM), and upstream infrastructure. Marginal changes in TPUs are now outstripping GPUs.
August 17: That Turning Point
“The Fed stepped up and chose to support growth. At that moment, I completed my pivot from extreme pessimism to aggressively adding positions. This macro environment felt exactly like Q3 2018—as long as growth can keep up, valuations can be digested.”
A TPU Cloud?
Google has been making TPUs for ten years. It’s not a new thing, but will it be the cloud hanging over Nvidia in 2026? Right now, they look more like efficient supplements for specific workloads.
The valuation ceiling for the Nasdaq 100 has been oscillating between 25 and 27.4 times. We are currently at the bottom of this range.
QQQ Valuation Path and Trading Range (2024-2025)
Note: Data reflects market volatility sensitivity within the 25x-27.4x valuation range
“Bubbles don't pop on their own,
unlessconvictionstarts to crumble.”
Next Chapter: Investment Logic & Framework—How to Distinguish Narrative from Fundamentals?
Logic & Framework
Don't get kidnapped by the story:
NarrativeandFundamentals: The line between life and death
When we were reviewing 2025 just now, our heads were full of the regret of "missing out." But honestly, if we did it all over again without an underlying logical framework, we’d likely fall into the same trap. The current market is half fiery narrative and half ice-cold fundamentals.
“Narrative is discounting the future; fundamentals are rooting in the present.”
I’ve been thinking about why some very "airy" AI stocks can skyrocket while giants like Google, holding a great hand, are constantly questioned. The core lies in whether you are investing in the "Narrative Phase" or the "Delivery Phase".
Narrative investing is about the "power of belief." The logic doesn't need to be a closed loop; it just needs a massive, unfalsifiable space for imagination. Once you enter fundamental scrutiny, the market starts doing the math: Can your model lead translate into gross margin? Is your ecosystem position exclusive?
My criteria:
- Narrative looks at the "ceiling," fundamentals look at the "moat"
- Narrative looks at traffic growth, fundamentals look at retention costs
- Narrative looks at the CEO's pitch deck, fundamentals look at the CFO's balance sheet
#Hot Take
“80% of research reports these days read like science fiction. True value discovery usually begins when the novel ends and everyone thinks it's boring.”
2026 Macro Outlook:
Growth Divergence
Fiscal support is no longer a blanket policy but targeted drip irrigation. Whoever is closer to compute and closer to policy is the winner.
Growth Expectation Divergence by Sector (2026E)
The Obsession with Google
Many say Google is getting old, but I look at its path to profitability. Its model leadership is a fact, but the market worries about its "left hand cannibalizing the right"—will the foundation of search advertising be undermined by AI? This valuation discount is essentially a fear of the "Innovator's Dilemma."
GPU vs ASIC: The Battle of Ecosystems
Nvidia's GPUs are indeed hard currency, but have you noticed? Big tech companies are all building ASICs (Application-Specific Integrated Circuits). It's not just about saving money; it's the endgame of vertical integration. Whoever can deeply bind algorithms and chips will survive the cost war after 2026.
Coming up Next
Will the AI bubble burst? Money is cheap; conviction is what's valuable
—— OpenAI IPO Preview: Profitability Challenges and Ecosystem Positioning →
Money is cheap,
conviction
信仰is what's truly valuable.
Now that we've finished talking about the 'hard-core' power play between GPUs and ASICs, we need to talk about something more abstract—valuations, bubbles, and this century's gamble on the future.
I'm often asked one question: Is the AI bubble getting too big? My answer is always: If there isn't a bubble, that's when it’s truly dangerous. The current market logic is actually quite simple: high-quality assets are too scarce globally, and money is searching everywhere for an exit.
When we talk about Nvidia’s P/E ratio, we’re actually talking about a kind of 'certainty.' But when that certainty spreads to the yet-to-be-listed OpenAI or SpaceX, things get interesting. This isn't just a simple discounted cash flow; it’s a premium on 'Artificial General Intelligence (AGI)' and 'interstellar civilization.'
“In this era, profits are for mortals, whereas ‘the right to interpret the narrative’ is for the gods.”
OpenAI IPO Outlook
Profitability Challenges and Ecosystem Positioning
OpenAI is in a delicate position right now. It’s like a unicorn born with a silver spoon—taking massive funding from Microsoft on one hand, while facing the immense pressure of a high burn rate on the other. Its valuation isn't based on how many ChatGPT subscriptions it sells, but on whether it can become the 'operating system' of the AI era.
SpaceX Imagination
Space Data Centers
The story Musk tells feels more like a 'dimensionality reduction strike.' If Starlink is turned into a distributed server in space, bypassing surface-level energy and cooling constraints, the entire valuation logic changes.
Market Sentiment Index: AI Faith vs. Macro Reality
Data Source: Podcast Deep Research Team Forecast
Potential Black Swan: What if faith wavers?
“My biggest worry is: when these AI companies go to raise funds, what happens if no one is willing to take the next round, or if everyone realizes the Scaling Law has hit a wall?”
“Then that’s a classic 'Lehman moment.' Current AI faith is built on the assumption that 'models will keep getting infinitely stronger.' Once that assumption loosens, the valuation of the entire chain—from chips to cloud services—will collapse.”
“So you're saying it's hope, not cash flow, supporting stock prices right now. Once hope shatters, money becomes worthless.”
Final Take
Don't look for
asafe harbor
inside the bubble. When the bubble pops, only those who hold core assets will survive. Since the macro environment is unavoidable, why don't we look at which neglected 'traditional fortresses' still exist in this crazy market?
Audience Q&A: Are there opportunities in traditional sectors? →
Stepping outside AI,
looking for the market's remaining warmth
After discussing SpaceX’s interstellar horizons and AI’s massive volatility, everyone’s concerns have become more pragmatic: If we look past Big Tech, is there any hope for traditional industries? How does the 2026 script play out? And where exactly are we currently standing in the cycle?
QA 01: Are there still opportunities in traditional sectors?
Energy and the Grid: The 'Endgame' of AI
Frankly, I don't see energy as a 'traditional industry' anymore. The endgame of AI is electricity; that's the consensus. While all the compute centers are in a cutthroat race, the transformers, copper cables, and nuclear power that can provide stable supply are essentially 'AI overflow beneficiary stocks.' This logic will shift from expectation to profit after 2025.
Finance and Infrastructure
If a downward interest rate cycle is established, those utility sectors that have been suppressed by high interest rates for three years will see a significant valuation recovery. This isn't a victory for growth; it's the return of the capital safe harbor.
“Don't forget the old master repairing the substation while everyone else is rushing toward supercomputers.”
The Inverse Investment of Consumption Downgrading
I'm bullish on sectors that provide 'cheap dopamine.' Under macro uncertainty, the lipstick effect still holds true, except this time it might be cheaper digital entertainment or high-value-for-money fast consumption.
QA 02: In 2026, what’s left of AI?
Many listeners have asked me: if 2024 is the year of infrastructure and 2025 is the year of applications, then what about 2026? My prediction is:The Great Shakeout Year.
By 2026, the market will no longer give you a premium just because you've 'integrated a certain large model.' At that point, the core of investment will lie in“the return of the moat”. We need to look for those companies that successfully transform AI into 'exclusive data assets.'
“We are not looking for 'AI companies,' but 'industry giants armed to the teeth by AI.'”
The current logic is to see who is buying GPUs; the logic in 2026 will be to see who is saving money and who is increasing revenue. I will focus on the penetration rate of AI Agents in the B2B sector. If a company's HR and financial systems haven't been completely restructured by AI, it probably won't be on my watchlist anymore.
QA 03: Cycle Positioning
Where are we in the cycle? I think it is“the eve of disillusionment”and“the mid-game of the paradigm shift”intertwined.
Frenzy Phase:
As long as it had an AI suffix, it rose; 100% sentiment premium.
Divergence Phase:
Skepticism over excessive infrastructure investment surfaces, and macro pressures begin to show.
Convergence Phase:
Fake demand goes bust, and the real winners take all.
Market Sentiment vs. Actual Application Implementation (Simulated Trendline)
Summary Sentiment
“Don't look for security in the bubble,
look for vitality in the wasteland.”
Having answered these questions, I think everyone now has a clear sense of the current situation.
Finally, let's talk about the final note of this episode—on the self-cultivation of an investor.
Related Episodes
Tesla vs. Waymo: The High-Stakes Battle for Self-Driving Supremacy
AI 2025-2026: Consensus, Conflict, and the Next Frontier of Scaling Laws
e220-hyrox-4z8s5
The €14 Billion Hermès Mystery: Heirs, Gardeners, and the War for Luxury
e218-netflix-b97ly