Rebought 80% of my original $SNDK long position at $171.31 ahead of $STX earnings last night.

Expected a beat from STX, given how the rest of the AI infrastructure names have performed so far this Q3.

SK Hynix also in with very positive commentary for memory

“DRAM and NAND customers are secured for the entire of 2026”

“… supply expected to remain relatively tight through 2027”

Good print for $TTMI with double beats coming in above the high-end of guidance. Low end of Q4'25 guide also above consensus. Stock +7% AH to $63.53. Actually surprised stock is not up more; may have been priced in by buy-side or possibly because the stock is very under-covered.

DC revenue was a key driver for rev in this quarter, inflecting higher at +45% y/y. In line with positive readthroughs from hyperscalers' earnings which indicated further acceleration in capex (supply constraints flagged by MSFT and GOOG!! META reports that they are "AI-compute starved"). Another positive is a shift towards AI accelerators and a strong push by GOOG for their TPUs; inline with our custom silicon thesis. A&D segment continues to be strong too, not much worries there.

There's also a very clear margin expansion story here which I think may be underappreciated by the market — which still perceives the company as a low-end commoditised PCB manufacturer. Q3 adj EBIT% at 12.3% is already higher than the midpoint of their LT target set in their 2023 analyst day. The company has been pushing aggressively into high-growth/margin sectors like high-end DCAI/networking and complex systems extending beyond PCB for A&D, and divesting low margin segments like mobile PCB. As the only high-end PCB manufacturer based in US, their NA factories allow them to provide turnkey high-end PCB with short lead times for hyperscalers that want to design custom silicon — AKA "premium revenue" with 1.5x - 2x ASP, which directly flows through to margins.

Forgot to mention but shorted $CVNA @ $362.92 over the past 2 days. Mainly just fading the mean reversion after the regional banks and auto-loans hoo-ha.

Stock is down -9% AH after their print; strong beat on sales but EBITDA% shockingly decelerated — probably what lead to the sell-off. Will dive in deeper later.

Current short basket includes $FIVE @ ~$145 avg, $CAVA @ ~$64.5 avg, $CRCL @ ~$127 avg, $SHAK @ $92.31 avg, $DKNG @ $33.9 avg, $CVNA @ $362.92 avg, $COIN @ $347 avg.

Solid print for $GOOG -- capex guide revised upwards, but that's fine as cloud backlog +46% SEQUENTIALLY! GOOG can continue spending as long as enterprise AI demand keeps up. Gemini MAUs also crossed 650mn, quite the exponential spike from Jul's 450mn. AI workload demand is just off the charts. Market is finally pricing GOOG as a dominant full-stack AI provider. As a bonus Youtube and Search also resilient, not bad! RIP GOOG bears

$CMG -17% in pre-market after warning that a large % of their consumers — specifically 25-35yo & consumers w HHI <$100k — are pulling back spend because of economy concerns. In line with our parallel economy thesis; S&P500 new highs, and yet consumers suffer.

Honestly don’t see how “premium” fast casual players like $SHAK and $CAVA can work here, given the overlap in demographics and pull back in spend. What more at such lofty valuations..? 58x/108x NTM p/e for SHAK and CAVA respectively.. Restaurants are known to have high operating leverage but that works both ways!

Speaking from my tastebuds, CAVA is basically a worse-tasting yet pricier CMG.. SHAK is actually nice but again, who wants to pay $20 for a burger set in this climate?

SHAK reports later, CAVA next week.

Added $CORZ @ from $20.25-$20.96 last night as shareholders voted and rejected CoreWeave’s acquisition offer for Core Scientific.

Closing $CRCL short @ $125 (+1% gain)

Still remain bearish on the company but borrow fees is just too expensive to keep the short on for long.

Closed 75% of $CAVA short @ $52.63 for +18% gain

Don’t want to take the print risk on this. Gut feeling; I think it should have traded down more on $CMG print but it didn’t. Also maybe expectations are already very low for this Q, so it may miss and still print up post earnings.

Shorted $HOOD @ $143.7
Shorted $PLTR @ $192.71

Again, just to hedge in case this momentum unwind gets worse

Closed remaining 25% $CAVA short @ $46.49

Might be closing more shorts later to go “longer”

Closed 50% of $SHAK short @ $95.4

Closed 30% of shorts across the board for $COIN @ $313.5, $HOOD @ $139, $PLTR @ $190

My manner of BTD

Added $SNDK long @ $192.45

Closed longs $763.HK @ $31.7 HKD and $$6869.HK @ $33.94 HKD, for -23% and -15% loss respectively

Quite a large % drawdown for these 2 trades, but thankfully sized these as small moonshots. Consolidating portfolio into core positions.

Shorting a small $SNAP position into print @ $7.4

Longed $COHR @ $134.73 on $LITE readthrough

Trimmed 40% of $FIVE short @ $160.25 for -10.5% loss.

Seems like a high probability Supreme Court is going to rule against Trump’s tariffs. Also there’s many other better short candidates, $FIVE might benefit from trade-down if pressures on consumer spend worsen. I still think the biz model is crap though.

Out of short btw, took a -20% cut on a mini position.

Perplexity deal is nuts, EBITDA flow-through on that deal amount for a $12bn eqv company is material — also not discounting the fact that there may be more on the way.