0) Theme of the Week — "The Semiconductor Singularity vs. The Software Apocalypse"
This week's tape is a monoculture cosplaying as a market. Nineteen YouTubers, one trade. The entire creator ecosystem is basically chanting "AMD, Nvidia, AI infrastructure, memory" in slightly different accents while quietly admitting they're trimming. Meanwhile, software (Service Now, Adobe, Shopify) is being thrown out with the bathwater because someone said the words "SaaS apocalypse" out loud, and apparently we're all six years old.
Underneath all that: the S&P just had a 6-week green streak, AMD ripped from $80 to $421 in a year, and every analyst on Wall Street is doing the exact thing they always do — chasing the price target up after the move. Two of the more honest sources (Everything Money, Couch Investor) are quietly saying: "yeah, this is starting to feel toppy." Most of the rest are saying "buy more, this is just the start" while in the same breath admitting they're net sellers. Cute.
1) What Actually Matters This Week
Strip away the YouTube hype-faces and the courses being sold. Three things are still live and unresolved:
a) The AMD post-earnings chase. AMD reported, Lisa Sue more than doubled the server-CPU TAM to $120B by 2030, analysts are now playing leapfrog with price targets ($400 → $530+). Stock closed ~$421. The question isn't "is AMD a good company" — it's "are you buying after a 5x in 12 months because Wall Street is panic-upgrading?"
b) The software bifurcation. Service Now (-58% from highs) and Shopify reported solid numbers. Market doesn't care. Why this matters: if the SaaS Apocalypse narrative breaks (which the actual numbers already suggest it has), this is a setup. If it holds, this is a value trap.
c) NVDA earnings on May 20. Roughly 10 days out. Two sources flagged the historical 2–4 week pre-earnings runup pattern. Whether you believe in seasonality or not, this is a known catalyst with a hard date.
Everything else — Bitcoin $80k chart talk, Iran ceasefire chatter, Marcado Libre's 13% drop — is noise relative to these three.
2) Consensus Themes Across Sources
Where they actually agree (and whether the agreement still holds):
Consensus 1: AI infrastructure / semis still have a runway. Cited by Chris Sain, Couch Investor, Larry Jones, Jeremy, Adriconomics, Zip Trader, MarketBeat (Hughes), Fired Up Wealth. Names: AMD, NVDA, Micron, TSM, Intel, Bloom Energy, Astera Labs. Still valid? Probably yes on a 12-month view. But every single one of these creators also admits the moves have been "insane" and several are trimming. When everyone agrees, the easy money is gone.That's not bearish, it's just real.
Consensus 2: Memory cycle (Micron, SanDisk, WDC) is overheated. Couch Investor, Jeremy, Larry Jones all flag it. SanDisk +3,660% in a year is the anchor stat. Still valid? Yes. Even the bulls are saying "don't chase here."
Consensus 3: AMD's CPU-to-GPU ratio shift is structurally bullish. Jeremy, Zip Trader, Chris Sain, MarketBeat all hit this. The 1:4/1:8 → 1:1 ratio for inference workloads is real and it's the most defensible part of the AMD bull case.
Consensus 4: Marcado Libre (MELI) is a long-term winner getting hated. Couch Investor, Fired Up Wealth, Asymmetric Investing. 29 quarters of 30%+ revenue growth, Q1 just printed +49%. Still valid? Yes, and arguably the cleanest "ignored quality" idea in the file.
Consensus 5: Meta is undervalued relative to its CapEx-cycle peers. Couch Investor, Adriconomics, Jeremy, Fired Up Wealth. Still valid? Yes, but this has been "the consensus contrarian play" for months. Crowd is small but loud.
Where the consensus is suspicious: when nine creators agree a stock is a buy and they all post videos in the same week, the trade is already crowded. Apply a mental haircut to every one of these.
3) Where Sources Disagree (and why it matters)
Disagreement 1: Are we in 2021 again?
Jeremy: emphatically NO. Bitcoin -35%, real estate flat, Dow components 5–87% from highs, money market funds bloated. Says it's "nothing like 2021."
Everything Money: emphatically YES on valuation. Buffett indicator at 132% above fair value, Shiller PE at 39+. Says "the next 10–15 years will not be good."
Adriconomics / Tom Nash / Zip Trader: somewhere in between. Concentrated euphoria in semis, broad market merely "expensive."
Who's right? They're describing different things. Jeremy is right on breadth/sentiment. Everything Money is right on aggregate valuation math. Both can be true. The trade-off: if you're a 30-year compounder, ignore. If you have a 3-year horizon, listen to Everything Money.
Disagreement 2: Oracle — buy or train wreck?
MarketBeat (Likenfeld, May 7-ish): SELL. $250B off-balance-sheet obligations, $25B/yr free cash flow negative, dependent on OpenAI revenue, Larry Ellison's Paramount backstop wildcard.
Zip Trader / Jeremy (implicitly): bullish on AI infra → Oracle is part of that.
Tension: Likenfeld's specific case is granular and falsifiable. The bull case is "AI go up." Likenfeld's argument is uncomfortable but specific. Take it seriously.
Disagreement 3: Service Now — falling knife or generational buy?
Zip Trader / Fired Up Wealth: GENERATIONAL BUY. -58% from highs, growing 21%, $4.6B FCF, 50%+ of new biz already on non-seat pricing (kills the SaaS apocalypse thesis for this specific name).
Almost everyone else: silent. Which itself is interesting.
Resolution: the Service Now bull case is the single most fundamentally specific argument in the entire file. Whether it plays out is a separate question.
Disagreement 4: Iran (the company, IREN).
Couch Investor: SKEPTICAL. They blew the $500M ARR Q1 deadline. Disclaimers in fine print are red flags. The Nvidia "deal" is Nvidia getting cheap warrants, not validation.
Kenen Grace / others: bullish, lumping it with the Neocloud trade.
Resolution: Couch's specific critique is well-documented. Iran has a credibility problem on guidance.
Already resolved disagreements (= stale debates):
Bitcoin "candle close above 80,400" (Stocks with Josh): the move it discusses is already a week old. Discard the specific levels.
Larry Jones' "Iran ceasefire deal" pump: already priced in or already faded. Stale.
"Sell in May" (MarketBeat early May): the source itself dismissed this thesis. Resolved — don't act on seasonality alone.
4) Refined Trade Hypotheses — WITH TIME LABELS
Reminder: this is probabilistic scenario planning, not financial advice. Every one of these can go wrong.
⚡ ACT TODAY / This Coming Week
TRADE 1: NVDA pre-earnings setup
Ticker: NVDA
Thesis: Historical 2–4 week pre-earnings runup pattern; earnings May 20.
Time label: 📅 THIS WEEK (window: now through ~May 18)
Trigger: Already in the window. Entry on dips below ~$205 if available.
Invalidation: NVDA breaking below the $190–195 zone with volume; broad market rolling over (SPY losing 729 per Stocks with Josh).
Conviction: Medium. The pattern is real but well-known, which dilutes the edge.
Source freshness: 🟡 (Kenen Grace's NVDA video, ~3 days old)
TRADE 2: SPY trend-trade with stop
Ticker: SPY / QQQ
Thesis: 6 straight green weeks; institutional money hedging via calls/puts but not selling. Trend intact above 729; broken below.
Time label: ⚡ ACT TODAY (intraweek setup)
Trigger: Hold above 729; reject if breaks below.
Invalidation: Daily close below 729 → likely retest 720.
Conviction: Low–Medium. Late in a 6-week run is the worst time to enter a trend trade.
Source freshness: 🟢 (Stocks with Josh, late Friday)
📅 THIS WEEK
TRADE 3: MELI post-earnings re-add
Ticker: MELI
Thesis: 29 consecutive quarters of >30% revenue growth, Q1 +49%, getting punished for strategic margin compression. Classic Mr. Market mood swing.
Time label: 📅 THIS WEEK (post-print weakness)
Trigger: Stock at $1,852 going into print last week. Buy zone per Fired Up Wealth: ≤$1,750.
Invalidation: Real evidence Amazon LATAM is taking durable share (not yet present); take rates collapse instead of stabilize.
Conviction: Medium-High. Best "ignored quality" name in the file.
Source freshness: 🟡 (Couch Investor, Fired Up Wealth, Asymmetric — all within 4 days)
TRADE 4: Service Now contrarian long
Ticker: NOW
Thesis: -58% from highs, 21% revenue growth, 50%+ of new business on non-seat (consumption-based) pricing — meaning the SaaS apocalypse thesis is already structurally addressed. Market hasn't repriced.
Time label: 📅 THIS WEEK / 🗓️ THIS MONTH (slow grind setup)
Trigger: Buy below $100; DCA zone $90–100. Earnings already reported; no immediate catalyst, which is part of the asymmetry.
Invalidation: CRPO growth slows below 20%; major Fortune 500 customer churn announcement; another aggressive guidance cut.
Conviction: Medium-High on the fundamentals; Low-Medium on timing (could chop for months).
Source freshness: 🟡 (Zip Trader, Fired Up Wealth, both within ~4 days)
🗓️ THIS MONTH
TRADE 5: AMD position management (NOT initiation)
Ticker: AMD
Thesis: Structural CPU TAM expansion is real. But after a 5x in 12 months and Wall Street panic-upgrading to $530+, this is a "trim and let runners run" situation, not a fresh entry.
Time label: 🗓️ THIS MONTH
Trigger (for existing holders): Trim 10–20% on continued strength; re-add on >15% pullback.
Trigger (for non-holders): Wait. Don't chase 5-baggers post-Wall-Street-upgrade-spree.
Invalidation of the long-term thesis: GPU revenue growth deceleration in the back half (currently accelerating); Lisa Sue guidance cut.
Conviction: High on the company, Low on the entry price for new positions.
Source freshness: 🟢🟡 (Jeremy, MarketBeat, Zip Trader, Adriconomics — all within 5 days)
TRADE 6: Meta on weakness
Ticker: META
Thesis: CapEx cycle compressing free cash flow now → flips when the spend rate stabilizes. Same playbook that worked in 2022–2023.
Time label: 🗓️ THIS MONTH
Trigger: Continue accumulating sub-$700, heavier sub-$650.
Invalidation: Genuine ad-revenue weakness (not yet visible); CapEx growth re-accelerating beyond 2025 levels.
Conviction: Medium-High.
Source freshness: 🟡 (multiple sources within the week)
TRADE 7: Oracle short / avoid (contrarian to AI consensus)
Ticker: ORCL
Thesis (Likenfeld's case, not endorsed without scrutiny): $250B off-balance-sheet obligations, FCF-negative ~$25B/yr, dependent on OpenAI's ability to pay $300B+ over 5 years, Ellison's Paramount backstop is a tail risk.
Time label: 🗓️ THIS MONTH+ (slow-burn thesis)
Trigger: No specific entry; this is a "don't own it" call with optional hedge structures for the brave.
Invalidation: OpenAI hits its revenue projections and Oracle starts converting backlog smoothly through 2026.
Conviction: Low-Medium. The case is specific but the timing is unknown.
Source freshness: 🟡 (MarketBeat / Likenfeld)
🔴 EXPLICITLY DISCARDED (stale or already played out)
Bitcoin 80,400 candle-close trade (Stocks with Josh) — 🔴 STALE. The exact level discussed is more than a week old; price has moved. Discard the specific trigger.
"Sell in May" seasonality — even the source (MarketBeat) said don't act on it. Done.
Iran ceasefire pump trade (Larry Jones) — already in the tape; the news is fully priced.
APLD at $5.44 (Chris Sain reminiscing) — that ship sailed in 2023. Not a current trade idea, just a flex.
5) Risk Map
What could break the overall thesis, on what timeline:
Within 1 week:
SPY losing 729 → quick mechanical 1–2% pullback (Stocks with Josh's level).
NVDA earnings May 20 disappoint — would crater the entire AI-infra basket including AMD.
Geopolitical escalation (Middle East) re-emerges.
Within 1 month:
Memory pullback (SanDisk, MU, WDC) cracks the broader semi narrative; AMD/NVDA decouple or get dragged.
Hyperscaler CapEx guidance for H2 2026 disappoints (Q2 earnings).
A single high-profile Neocloud (CRWV, IREN, NBIS) misses badly → narrative contagion.
Within 1 quarter:
Q2 earnings season (July) where the 27% YoY EPS growth either repeats or collapses.
Fed policy shift if inflation re-accelerates.
Within 1 year:
The Buffett-indicator/Shiller-PE math (Everything Money) finally meeting reality. If we're 132% above fair value, mean reversion is a question of when, not if.
Oracle scenario plays out (OpenAI can't pay) → systemic AI infrastructure repricing.
Real revenue-growth deceleration in AMD (Jeremy's "1–2 year" warning bell).
6) If We're Wrong, We're Wrong Because…
Honest post-mortem, focused on timing because that's what kills people:
1. Wrong on duration of the AI cycle. Paul Tudor Jones thinks AI has 1–2 more years. Every creator in this file is positioned for at least that. If the cycle is shorter — say it tops in summer 2026 — every "🗓️ THIS MONTH" trade above is wrong by months. The most likely failure mode for AMD/NVDA isn't that the businesses break, it's that the valuations peak before the fundamentals peak (Jeremy's own "stocks price in years ahead" point, used against him).
2. Wrong on the SaaS apocalypse. Everyone in the file dismisses it (or counters it for specific names like Service Now). What if it's real but slower? Service Now's "consumption pricing" is great in theory; in practice, customers buying fewer seats is a transition, not a clean swap. The bull case requires consumption growth to more than offset seat decline. If it merely matches it, you get a value trap.
3. Wrong on Meta/MELI patience. Both names have looked "cheap" for multiple quarters. The thesis is that the market eventually rewards them. The market's rewards are not on a schedule. Could chop sideways another 6 months while opportunity cost mounts.
4. Wrong on Service Now's bottom. Down 58% from highs feels like a bottom. -58% became -70% for a lot of stocks in 2022. There's no rule that says it can't get worse before it gets better.
5. Wrong on the broad valuation reckoning. Everything Money's case is uncomfortable but mathematically grounded. If they're right on the 10-year math, every long position sized for the next decade is wrong-sized today. The way to be wrong on timing here is to assume mean reversion happens "soon." It might not happen for years.
6. Wrong because too many YouTubers agree. This is the meta-meta point. When 19 out of 19 creators are bullish on AI infrastructure (with token nervousness), the marginal buyer is exhausted. Markets top when there's no one left to convert. We may be closer to that than the consensus admits.
Not financial advice. Probabilistic thinking, scenario planning, and risk management only. The market does not care about your thesis, your YouTuber, or your feelings. Size accordingly.