Behind the video
Don't Buy the Hype! Investing Lessons from Past Bubbles
Overall Thesis
Using historical FOMO cycles like Plug Power, Lucid, Amazon, and Cisco, the host argues today's hottest AI-era stocks — Nvidia, Micron, Lam Research, and Bloom Energy — show the same inflated-expectations pattern that preceded past crashes. He says he is avoiding these names until the FOMO subsides and prefers to buy proven winners once valuations reset lower.
Narratives
NVDANvidiaNvidia shows classic peak-of-inflated-expectations traits: explosive share gains, lack of profitability relative to its scale, heavy planned capex versus thin cash cushion, and valuation multiples reminiscent of dot-com-era stocks that later crashed. The host views it as very high-risk and is avoiding it at current prices.
Key Arguments
- Shares up over 10x in 3 years and 245% in 5 years on AI demand
- Not profitable despite a reported $59 billion market cap figure cited by the host
- Planning $25 billion of capex against only about $9 billion cash and $8.4 billion debt
- Forward EV/sales of 8x echoes multiples (and higher) seen at dot-com-era companies before they crashed 80-99%
Risks acknowledged
- Company has a large backlog of demand
- Is aggressively building out data centers amid genuine AI adoption
Predictions (1)
MUMicron TechnologyMicron's current AI-driven memory boom echoes its historical boom-bust cycles, where overbuilding capacity previously led to heavy losses (including as recently as 2023). The host doubts this time is different and is avoiding the stock at its current $1.1 trillion valuation.
Key Arguments
- Memory industry has a history of overbuilding capacity followed by steep losses
- Micron lost money as recently as 2023 before the current AI-driven rebound
- Expects capacity to expand and users/developers to become more memory-efficient, undermining current demand surge
- Cites recent Google-related developments as an early sign of changing memory efficiency dynamics
Risks acknowledged
- Acknowledges the current AI-driven demand explosion for memory is real
Predictions (1)
Lam Research makes important semiconductor equipment, but its valuation (18x EV/sales against only a 13% 10-year revenue CAGR) and 400% share price run since April 2025 represent the peak of inflated expectations. The host explicitly says he will not buy the stock here.
Key Arguments
- EV/sales of 18x versus a 10-year revenue CAGR of only 13%
- Shares up 400% since April 2025
- Views the current price as the peak of FOMO-driven inflated expectations
Risks acknowledged
- Acknowledges Lam Research makes very important semiconductor manufacturing equipment
Predictions (1)
BEBloom EnergyBloom Energy's roughly 28x rally since 2024 and 30x sales valuation reflect a short-term power-scarcity boom rather than a durable long-term business, especially since most of its output still relies on natural gas rather than the hydrogen future it markets. The host views the $74 billion valuation as unsustainable long-term.
Key Arguments
- Shares up about 28x since 2024 on desperate near-term power demand
- Only recently became operating-profit positive
- Trading at 30x sales for what the host views as a short-term solution
- Mostly uses natural gas rather than hydrogen despite marketing a hydrogen future
- Expects more efficient, lower-cost power infrastructure to erode current pricing power over the next decade
Risks acknowledged
- Currently has essentially unlimited short-term demand and pricing power from power-starved customers
- Host states he would 'love for that hydrogen future to be a reality'
Predictions (1)
Used as a historical case study: Plug Power went through repeated FOMO cycles (IPO era, mid-2010s hydrogen hype, 2021 pandemic hype) where share surges were never matched by a path to profitability, and each time shares eventually fell 80-90%+.
Key Arguments
- Revenue grew substantially during hype periods but losses widened rather than narrowed
- Shares have fallen roughly 80-90%+ after each of three separate hype cycles
LCIDLucid GroupCited as a historical example of story-driven FOMO: Lucid's SPAC-era shares went from a peak of $511 (adjusted) to a 99% decline, illustrating that long-term holders of hype-driven stocks can be wiped out even without a hard current call on the stock.
Key Arguments
- Shares fell roughly 99% from their SPAC-era adjusted peak of $511
AMZNAmazonUsed as a historical example showing even phenomenal long-term winners can be devastating for investors who buy at the peak: Amazon fell about 93% after its 1999 peak and did not return to breakeven for buyers until late 2009.
Key Arguments
- Shares fell about 93% after the 1999 peak
- Took until late 2009 for 1999 buyers to break even
Risks acknowledged
- Amazon ultimately became one of the great long-term winners of the dot-com era
Used as a historical example of dot-com-era hype: Cisco shares rose roughly 75,000% from 1990-2000 before crashing about 90%, and only recently regained its prior valuation despite remaining a large, profitable company.
Key Arguments
- Shares rose roughly 75,000% from 1990 to the 2000 peak
- Fell about 90% from peak and only recently regained that valuation level
- Remained profitable and large throughout, unlike weaker dot-com peers
Hedges & Caveats
- Acknowledges some past FOMO stocks like Amazon eventually became great long-term winners after crashing
- Notes AI demand and backlog for Nvidia and memory makers are real, not fabricated
- States bubbles bursting doesn't mean all companies in the space disappear
- Admits he previously owned Bloom Energy and 'should have held on to it'