| Ticker | Name | Research | View | What he said | At |
|---|---|---|---|---|---|
| GLD | SPDR Gold Shares | QT · SA · STK | Positive | Uses GLD flows as the capitulation gauge: 110 tons out in six months, no retail participation vs the 2011-12 top; with futures open interest at a 13-yr low and central banks (China accelerating) still buying, he sees bottoming/capitulation signs and has started nibbling miners — but warns of a possible "whoosh down" in an AI bust. | 59:36 |
| MSFT | Microsoft | QT · SA · STK · FA | Neutral | The survivor: "taking the right steps," offering low-cost model options (even considering DeepSeek) to protect its enterprise platform. But it will still "get smashed" in a 2000-style downturn (Microsoft fell 60% even as best performer) — stock already −20% YTD; then it becomes a buy. | 24:14 |
| GOOGL | Alphabet | QT · SA · STK · FA | Neutral | A survivor with real cash flow (search, YouTube) to fall back on — one of only two hyperscalers currently cash-flow positive. But depreciation expenses double (17%→35% by 2028), so the PE "adjusted for non-success" is much higher than the ~24× headline. | 20:01 |
| META | Meta Platforms | QT · SA · STK · FA | Neutral | Spending 100% of cash flow on data centers and just cut model prices 75% — but has a real cash-flow business to fall back on. The metaverse precedent: when Meta stopped burning $85B on the metaverse the stock rallied; he expects a spending-cut trigger to do that again. | 26:26 |
| 000660.KS | SK Hynix | STK | Neutral | Cited as an exhaustion signal, not a pick: its $27B equity raise last week (like SpaceX) is well-timed selling into the top — part of the shift from buybacks to "massive equity issues" that marks a market top. | 30:05 |
| 005930.KS | Samsung Electronics | STK | Neutral | Named among the memory/fab builders (with Micron and the Chinese CXMT/YMTC) racing to add semiconductor-fab capacity for data centers — the setup for a coming oversupply in DRAM/NAND. | 51:41 |
| BABA | Alibaba | QT · SA · STK · FA | Neutral | Maker of the Qwen model: a Vinod Khosla-backed startup reportedly shrank Qwen 3.6 to run on an iPhone 17 Pro — "you won't need the data center." Cited as an efficiency threat that should "put fright in Jensen Huang's dreams," not as a stock pick. | 41:30 |
| TSLA | Tesla | QT · SA · STK · FA | Neutral | Datapoint in the token-spending pullback: Elon has put a $200 weekly limit on employees' token spending — evidence even the biggest players are reining in AI costs. | 13:41 |
| COIN | Coinbase | QT · SA · STK · FA | Neutral | Another token-pullback datapoint: cutting its AI spending in half and shifting to lower-cost models after the "token maxing" surge. | 13:55 |
| DeepSeek | DeepSeek (private, China) | — | Neutral | The Chinese open-source model whose feared impact is now real: gets 90% of tasks done at ~1.5% of the cost — the pricing wedge forcing US frontier labs to cut prices; Microsoft is even considering using it. | 14:08 |
| Zhipu | Zhipu AI (private, China) | — | Neutral | Its latest model now exceeds the capabilities of Gemini 3.5/3.1 (per artificial-analysis rankings) — evidence the top Chinese models have caught the US frontier while costing a fraction. | 16:03 |
| Reflection | Reflection (private, US) | — | Neutral | A US-based startup using the same low-cost techniques as the Chinese modelers — his proof the pricing pressure survives even if Chinese models get banned; Microsoft is building its own low-price models the same way. | 39:10 |
| CXMT | ChangXin Memory Technologies (private, China) | — | Neutral | Chinese DRAM maker with a ~$5B IPO coming — part of the wave of memory-fab capacity (with Micron, Samsung, YMTC) that sets up a coming oversupply. | 51:04 |
| YMTC | Yangtze Memory Technologies (private, China) | — | Neutral | Chinese NAND maker with another offering coming — more new memory supply feeding the oversupply thesis. | 51:20 |
| NVDA | Nvidia | QT · SA · STK · FA | Negative | Poster child of the earnings bubble: its numbers soar as hyperscalers place GPU orders recognized as revenue immediately, while the buyers' costs (depreciated 5-6 yrs) lag — an inflated-earnings gap that collapses when spending slows. Efficiency breakthroughs (a model on an iPhone) threaten chip demand. | 6:20 |
| MU | Micron Technology | QT · SA · STK · FA | Negative | "As cyclical as they come" — had negative gross margins in 2023, now 85% and stock up 700% this year on the DRAM order surge. Unsustainable: talking about $250B of capex into a coming memory oversupply. | 5:51 |
| WDC | Western Digital | QT · SA · STK · FA | Negative | Another component maker booking data-center orders as immediate revenue — one of the supplier gains he stripped out to show S&P earnings growth was really single-digit. | 8:36 |
| STX | Seagate Technology | QT · SA · STK · FA | Negative | Named with Nvidia and Western Digital as the suppliers whose one-time gains he stripped from the S&P — earnings inflated by the buildout, exposed when it slows. | 8:36 |
| ORCL | Oracle | QT · SA · STK · FA | Negative | The most stretched: spending 100% of revenues on data centers, debt is 2.5× last year's sales, stock −28% YTD and −60% off its highs. He believes some of these debt-laden names end up going bankrupt. | 18:19 |
| AMZN | Amazon | QT · SA · STK · FA | Negative | Spending 100% of cash flows on data centers; got stuck with a $500M one-month token bill. Its recent $25B bond added 8bps to the 10-year and it now has to borrow in euros/yen/francs/pounds — it has saturated the US market, a sign of exhaustion. | 30:22 |
| SpaceX | SpaceX (private) | — | Negative | Its $85B IPO at a ~$2.2T valuation — the largest ever — is a top signal: no earnings, "all story," well-timed selling into the froth as the market shifts from buybacks to massive equity issuance. | 29:59 |
| OpenAI | OpenAI (private) | — | Negative | Still losing money in Q1 even amid the token surge; wants an IPO because it needs lots of cash. With Anthropic it represents ~50% of the hyperscalers' backlogs — at "great risk" as its pricing gets undercut; he expects some such names to go bankrupt. | 19:07 |
| Anthropic | Anthropic (private) | — | Negative | Like OpenAI: burning cash, wants an IPO, being forced to slash prices and offer subsidies as Chinese open-source models undercut it. Half the hyperscalers' backlogs sit with these two frontier labs — if they hit trouble it reflects straight back into the hyperscalers. | 19:07 |
"View" is Fred Hickey's stance in this conversation (Positive / Neutral / Negative), not a price rating. Many names are cited as evidence for the bubble/oversupply thesis rather than as trade ideas. Research links: QT Qualtrim · SA Seeking Alpha · STK Stock Analysis (omitted where no clean page exists). Private companies have no ticker.
A jargon-free summary of the thesis behind each name — what it actually is and why he holds that view. (Plain-language companion to the table above; renders on each ticker's consolidated page.)
GLD is the big gold ETF — a share is a claim on physical gold in a vault. Fred doesn't cite it as a trade so much as a thermometer for how washed-out gold sentiment is: 110 tons of gold have left GLD in six months, and unlike the 2011-12 top there was never a retail buying frenzy to unwind. On top of that, bets in the gold futures market have collapsed to a 13-year low, and a sentiment gauge called the BPGDM (the share of gold-mining stocks in an uptrend) crashed from 100% in January to 2%. To a contrarian, everyone already having sold is the setup for a bottom.
What's still quietly buying is central banks — China especially — who now hold more gold than US Treasuries and keep adding as a way to move off the dollar. That's why gold keeps defending $4,000. He's started "nibbling" on miners (Scotiabank says they're the cheapest in over four decades, ~10× earnings) but is holding back real buying because an AI crash could briefly drag gold down with everything else in a scramble for cash.
Microsoft is the name he most respects here: it's "taking the right steps" by offering customers cheaper AI options (even open-source Chinese models like DeepSeek if needed) so it keeps control of its enterprise platform rather than getting undercut. He's confident it survives the bust. But "survives" isn't "spared": in the 2000 crash the NASDAQ fell 83% and Microsoft — the best performer — still dropped 60%. The stock is already down 20% this year. His plan is to buy it after the washout, not now.
Google's saving grace is that its old business — search and YouTube ads — still gushes cash, so it has something real to fall back on when the AI spending stops paying off (it's one of only two hyperscalers still cash-flow positive after data-center costs). The catch is accounting: the huge sums spent on chips and buildings get expensed slowly over years ("depreciation"), and Google's depreciation roughly doubles by 2028 (from 17% to 35% of something). If AI revenue doesn't show up to cover it, real earnings sink — so the stock's "reasonable" ~24× earnings is much richer once you assume the buildout disappoints.
Meta is spending every dollar of cash flow on data centers and just cut its AI model prices 75% — a sign the economics are cracking. But like Google it has a real, cash-rich advertising business underneath. Fred's key point is a precedent: a few years ago Meta's stock got hammered while it burned $85 billion on the metaverse, and the moment it stopped spending, the stock took off. He thinks the first hyperscaler to admit "we have to cut AI capex" will get the same reward — and that admission is probably the trigger for the whole AI unwind.
SK Hynix is a giant Korean memory-chip maker, but Fred mentions it as a warning flag, not a pick. It just raised $27 billion in new stock last week. When companies rush to sell their own shares to the public (like SpaceX's IPO the same week), it usually means insiders think the timing is good — i.e. prices are near a top. The market has swung from companies buying back their shares to companies flooding the market with new shares, which is classic late-cycle behavior.
Alibaba makes Qwen, one of the strong open-source Chinese AI models. The eye-opener: a startup (Fred credits Vinod Khosla) reportedly shrank Qwen 3.6 down to run on an iPhone 17 Pro — meaning some AI tasks might not need a giant data center at all. That's a direct threat to the "we'll need endless chips and data centers" story, which is why he says it should scare Nvidia's Jensen Huang. He's flagging Alibaba as a source of that disruptive efficiency, not recommending the stock.
DeepSeek is a Chinese open-source AI model. A year ago people feared it; now the fear is reality: it reportedly does 90% of what you need at about 1.5% of the cost of the expensive US "frontier" models. Because it's so much cheaper, companies are switching to it in droves — which forces OpenAI and Anthropic to slash their own prices. Even Microsoft is reportedly considering using it. It's the wedge collapsing the pricing assumptions the whole AI buildout was justified on.
Zhipu is another Chinese AI lab, and its newest model now scores above Google's Gemini 3.5/3.1 on independent quality rankings. The point: the top Chinese models have caught up to the best US models while costing a fraction — so the idea that US labs can charge premium prices for superior AI is falling apart.
Reflection is a US startup using the same "do more with less" efficiency tricks as the Chinese labs. Fred cites it to make a point: even if the government banned Chinese AI models on national-security grounds, the price war wouldn't stop — American companies (Reflection, and Microsoft building its own low-cost models) can undercut the expensive frontier models the same way. Cheap AI is here to stay.
Nvidia is the clearest example of his "earnings bubble." When a hyperscaler orders GPUs, Nvidia books that as revenue and profit almost immediately, so its numbers look spectacular. But the customer only spreads the cost over 5-6 years (depreciation), so the buyer's pain is hidden while Nvidia's boom is front-loaded. When the spending slows — or efficiency breakthroughs (a model running on a phone) mean fewer chips are needed — that inflated-earnings gap snaps shut. In a 2000-style bust these supplier stocks fall hardest.
Micron makes memory chips (DRAM), a famously boom-and-bust business — it had negative gross margins as recently as 2023 (it lost money on every chip) and now enjoys 85% margins, sending the stock up 700% this year on the data-center order surge. Fred's warning is simply that this is the peak of a cycle, not a new normal: Micron is talking about $250 billion of new factory spending just as China's CXMT and YMTC and Samsung all add capacity — the recipe for a memory glut that crushes prices and margins.
Western Digital makes storage (hard drives and flash). It's in the same bucket as Nvidia and Micron: a component supplier booking data-center orders as instant revenue, so its earnings look inflated by the buildout. When Fred asked AI to strip these supplier gains out of the S&P's profits, the market's "28% growth" shrank to single digits — Western Digital was one of the names he removed.
Seagate, like Western Digital, is a storage maker riding the data-center order wave. He names it as one of the supplier gains that flatter S&P earnings today and would reverse when the buildout slows — earnings that look real now but are borrowed from a spending binge.
Oracle is the most dangerously stretched name in his view. It's plowing 100% of its revenue into building data centers and has piled on debt equal to 2.5× its annual sales. The stock is already down 28% this year and 60% from its highs. He thinks some of these heavily indebted AI-infrastructure companies actually go bankrupt if the revenue they're betting on doesn't arrive — and Oracle is his prime candidate.
Amazon is spending all of its cash flow on data centers and even got hit with a $500 million bill for one month of AI tokens. The tell he watches is in the bond market: Amazon's recent $25 billion bond was so large it nudged up the 10-year Treasury yield, and Amazon now has to borrow in euros, yen, Swiss francs and pounds because it has soaked up all the demand for its debt in the US. When a company has to scrape the bottom of every funding market, that's a sign of exhaustion near a top.
SpaceX's IPO — $85 billion raised at roughly a $2.2 trillion valuation, the largest ever — is Exhibit A of his "top signal." It makes no earnings; it's "all story." When the market shifts from companies buying back their own shares to insiders selling huge new share offerings at sky-high valuations, that's the kind of well-timed selling into froth that tends to mark the end of a bubble.
OpenAI (maker of ChatGPT) was still losing money in the first quarter even during the token-spending boom, and it wants to go public precisely because it constantly needs cash. The bigger risk: OpenAI and Anthropic together account for about half of the hyperscalers' order backlogs. So if cheaper Chinese models keep undercutting their pricing and they run into trouble — Fred thinks some AI companies go bankrupt — the damage feeds straight back into Microsoft, Amazon and the others counting on those orders.
Anthropic (maker of Claude) is in the same spot as OpenAI: burning cash, wanting an IPO to raise more, and being forced to slash prices and hand out subsidies as cheaper Chinese open-source models eat its lunch. Because these two frontier labs sit behind roughly half of the hyperscalers' backlog orders, their trouble is the hyperscalers' trouble — a single point of failure the market isn't pricing.
Summary & timestamps derived from the public YouTube video (transcript in transcript.txt) for personal study. Not investment advice. © Thoughtful Money / The High-Tech Strategist for source material.