15 years in U.S. equities. I watch the tape. The market usually tells you what it wants.
US
Joined July 2013
- Tweets74.6K
- Following956
- Followers2.1K
- Likes207K
The equity market is still pretending the AI infrastructure filter hasn’t tightened. Credit already has.
Hot inflation and rising hike odds are forcing a re-price of the debt-funded data-center model. $CRWV is the cleanest expression right now — funding costs are being re-rated, and the rest of the levered layer is only partially following.
Higher-quality, better-funded names are absorbing the pressure better. That’s why the real divide this week is balance sheet quality, not the headline index. $TSM remains the backbone on the funded side; further credit stress lower in the stack eventually shows up in $NVDA demand expectations.
If the gap between funded and levered names stays narrow into the close, strength in the weaker balance sheets is still temporary.
The market is ranking the AI capex trade by funding, not by narrative.
Still watching the tape.
Credit is already doing the work the equity tape is still pretending to ignore.
$CRWV bonds sitting near 15% yield is the cleanest signal that the debt-funded data-center model is being re-rated in real time. Higher-quality balance sheets are absorbing the pressure better — that’s why $TSM is holding up on the funded side while the levered layer lags.
If the gap between funded and levered names stays narrow into the close, any bounce in the weaker sheets still looks temporary. The market is ranking the AI capex trade by funding capacity, not narrative.
2
117
Credit tightened the filter earlier in the day. Equity only partially followed.
The higher-quality parts of the stack absorbed the pressure better, but the gap between funded and levered names didn’t widen cleanly into the close.
Friday’s open will show whether that gap starts to reassert, or whether the market is willing to give the weaker balance sheets another session of leeway.
If the gap stays narrow, treat any strength as temporary.
Still watching the tape.
2
5
167
Gus Fring retweeted
Dow led, Nasdaq lagged, but the split inside tech was the real story.
Quality software (MSFT) and some memory (MU) held up. NVDA, AMD, and the equipment names (AMAT, LRCX) didn’t. $AVGO was only mildly red — not strong, but not leading the downside either. $CRWV was weak early but recovered into the close. Healthcare and defensives were green.
The market is still sorting who can fund growth at these rates, not killing AI demand outright. But the distinction isn’t clean yet.
Still just watching the tape.
Dow led, Nasdaq lagged, but the split inside tech was the real story.
Quality software (MSFT) and some memory (MU) held up. NVDA, AMD, and the equipment names (AMAT, LRCX) didn’t. $AVGO was only mildly red — not strong, but not leading the downside either. $CRWV was weak early but recovered into the close. Healthcare and defensives were green.
The market is still sorting who can fund growth at these rates, not killing AI demand outright. But the distinction isn’t clean yet.
Still just watching the tape.
Yeah this tracks with what I’m seeing.
More of a leverage hangover than the AI trade dying.
Not all semis are the same though — server hardware getting smoked harder while the higher quality compute names are holding up better
dip has been bought.
$7500 in my solo 401k...idea here was to buy semis that are really getting hurt or names that I don't mind holding for a decade and are near their 52-week lows
semis getting hurt bad: $SMCI $TSM $AVGO $NBIS
out of these 4, SMCI is the one I like the least but it's also touching the 3month low and DELL has taken enough of a beating for SMCI to get a rebound if there is any bounce within server racks
the two large cap semis like $TSM and $AVGO I can see myself holding for a very long time as they are relatively safe and have real moats, $NBIS should bounce here but maybe more neocloud fears hurt it...the diversification of their business is some hedge to the pure GPU licensing business
the other 3 $TSLA $RKLB $SOFI i can imagine holding for a long time and 40% drawdowns feel like enough pain but I believe they all 3 will be fine in the long term, Tesla's short term catalyst is the roadster event, Rocketlab is down aggressively from $150 so I see that being a spot where some buyers step in even with the neutron delays, and $SOFI should benefit if yields come down but obviously rate hikes are an issue for them and every bank
didn't do much in my regular brokerage...bought most of the SaaS dips months ago with $PLTR $ZETA $SHOP $RDDT and have not been playing semis outside of $NVDA and $AVGO really...still holding the $DRAM $INTC and $BE position I bought in August which are either slightly up/flat since then
maybe we are headed another leg lower before the midterms but time will tell now 🤷♂️
2
11
274
Gus Fring retweeted
Everyone’s calling this the end of the AI trade.
I think it’s a leverage hangover — not proof the thesis is broken.
We’ve seen roughly a week of meaningful selling. Nasdaq is off its highs, semis have taken a harder hit.
The real question isn’t whether these stocks have fallen enough.
It’s whether earnings can catch up with expectations.
Watching:
$AMD
$SMCI
$WDC
$STX
$COHR
$LITE
$AAOI
$NRG
$VRT
$APLD
Three things matter next:
• Inflation and the path of rates
• AI capex vs. actual returns
• Earnings, margins, and forward guidance
If spending keeps growing but profits don’t follow, the market has a bigger problem.
No need to nail the bottom. Scale in selectively and let the numbers do the talking.
Everyone’s calling this the end of the AI trade.
I think it’s a leverage hangover — not proof the thesis is broken.
We’ve seen roughly a week of meaningful selling. Nasdaq is off its highs, semis have taken a harder hit.
The real question isn’t whether these stocks have fallen enough.
It’s whether earnings can catch up with expectations.
Watching:
$AMD
$SMCI
$WDC
$STX
$COHR
$LITE
$AAOI
$NRG
$VRT
$APLD
Three things matter next:
• Inflation and the path of rates
• AI capex vs. actual returns
• Earnings, margins, and forward guidance
If spending keeps growing but profits don’t follow, the market has a bigger problem.
No need to nail the bottom. Scale in selectively and let the numbers do the talking.
Gus Fring retweeted
The market isn’t killing AI demand.
It’s sorting who can still finance growth at higher rates.
Credit cracked first with $CRWV. Bonds near 15% told the story before the stock did. Debt-funded AI infrastructure is now being priced on the cost of capital, not just the backlog.
The rest of the stack is getting repriced unevenly: • Compute ($NVDA, $AVGO) holding up better • Interconnect and power getting dragged • Memory still under pressure from both rates and competition
Hot inflation data didn’t break the AI thesis. It just made financing more expensive, so relative strength and credit are doing the filtering.
Watch whether the pressure stays isolated to the leveraged names or starts bleeding into the higher-quality parts of the stack. The tape will show it.
The market isn’t killing AI demand.
It’s sorting who can still finance growth at higher rates.
Credit cracked first with $CRWV. Bonds near 15% told the story before the stock did. Debt-funded AI infrastructure is now being priced on the cost of capital, not just the backlog.
The rest of the stack is getting repriced unevenly: • Compute ($NVDA, $AVGO) holding up better • Interconnect and power getting dragged • Memory still under pressure from both rates and competition
Hot inflation data didn’t break the AI thesis. It just made financing more expensive, so relative strength and credit are doing the filtering.
Watch whether the pressure stays isolated to the leveraged names or starts bleeding into the higher-quality parts of the stack. The tape will show it.
Gus Fring retweeted
Hot CPI didn’t kill AI demand.
It just made financing AI growth more expensive.
$CRWV’s bonds hit 15% and the stock got smoked. That’s the signal. Debt-funded AI clouds are getting priced on the cost of money now, not just the backlog.
Interconnect and power got clearly dragged. Whole stack is getting repriced when the cost of capital rises.
Memory is still under pressure from the CXMT + rates story, but it wasn’t the weakest layer today.
Compute held up better. $NVDA and $AVGO are still red, but they held up better than most of the stack. Higher rates just make the multiples harder to defend.
Today wasn’t AI demand breaking. It was the market sorting who can still make the numbers work at higher financing costs. Credit cracked first.
Credit and relative strength matter more than the CPI headline
Hot CPI didn’t kill AI demand.
It just made financing AI growth more expensive.
$CRWV’s bonds hit 15% and the stock got smoked. That’s the signal. Debt-funded AI clouds are getting priced on the cost of money now, not just the backlog.
Interconnect and power got clearly dragged. Whole stack is getting repriced when the cost of capital rises.
Memory is still under pressure from the CXMT + rates story, but it wasn’t the weakest layer today.
Compute held up better. $NVDA and $AVGO are still red, but they held up better than most of the stack. Higher rates just make the multiples harder to defend.
Today wasn’t AI demand breaking. It was the market sorting who can still make the numbers work at higher financing costs. Credit cracked first.
Credit and relative strength matter more than the CPI headline