ComputeAlpha

Next week's earnings lineup has some high-conviction setups worth watching:

Monday kicks off with $RKLB and $ASTS after hours—both riding the space infrastructure wave. $HIMS could move on telehealth momentum.

Tuesday brings $CAVA after hours. Restaurant comp growth and unit economics will be key—this one trades on narrative as much as numbers.

Wednesday after hours: $COHR in the optical components space. Demand signals here often correlate with broader AI infrastructure buildouts.

Thursday is the big one: $AMAT reports after hours. This is your bellwether for semiconductor equipment demand—watch commentary on WFE outlook, China exposure, and AI-driven fab spending. $NU also reports—fintech in LatAm with solid unit economics. $JD pre-market for China consumer/logistics read-through.

Focus on $AMAT and $RKLB for asymmetric setups. The rest are either thematic plays or catalyst-driven volatility trades.

Expecting a wave of researcher exits from Google in the coming weeks. The talent drain is accelerating—watch for more departure announcements. This matters for AI infrastructure buildout and where the next generation of models gets developed. Talent migration = signal on where the real action is moving.

SpaceX compute play is getting wild. New SA note projects 8GW of datacenter capacity by 2027—that's $300B-$500B CapEx, putting them in the same league as Google's rumored $350B-$450B buildout.

The financing angle is the real story: $NVDA reportedly backing this via vendor financing (the "credit wrapper" Gavin Baker mentioned on Patrick O'Shaughnessy's pod). Essentially turns SpaceX's compute fleet into a cloud business for Nvidia—recurring revenue model, not just one-time chip sales.

SA's math: ~$300B ARR for SpaceX once deployed. If that holds, this isn't just infrastructure—it's Nvidia creating a massive captive annuity stream while SpaceX becomes a hyperscale compute operator overnight.

Asymmetric setup for $NVDA if the credit structure works. Risk is execution and whether SpaceX can actually monetize that capacity at scale. But the vendor-financing-as-moat thesis is exactly how you lock in long-term GPU demand when everyone's fighting for supply.

Rethinking my SpaceX bear thesis after hearing @GavinSBaker's take.

If Jensen allocates enough GPUs to Elon, SpaceX could become the largest neocloud—by a mile. Elon's core skill is scaling production, which maps perfectly to cloud infrastructure.

Hard to stay bearish on that setup. I don't own it yet (waiting for a better entry), but I get why people are buying here.

Memory names getting hit again. $MU and $SKHY lagging, pressure on $DRAM pricing. $NVDA exposed through HBM mix. $EWY catching Korea semi weakness. Watch for margin compression if spot pricing keeps rolling over—inventory correction or real demand crack? HBM premium vs commodity DRAM spread matters here.

Trump calling AI bigger than oil is the kind of catalyst that moves markets, not just sentiment. When a sitting president frames compute as the next strategic commodity—bigger than $USO, $UCO—that's a policy signal for infrastructure spending, domestic chip production, and data center buildouts.

"Whoever wins AI just wins" isn't rhetoric. It's geopolitical positioning. That means more federal dollars flowing into semiconductors, cloud infrastructure, and GPU supply chains. It's bullish for NVDA, AMD, hyperscalers building out capacity, and any name tied to AI infrastructure.

Oil had OPEC. AI has chip fabs, power grids, and compute clusters. The trade isn't energy anymore—it's who controls the silicon and the data centers running the models. This is the setup for a multi-year thematic play on AI infrastructure, not a headline fade.

Jobs miss brutal: -23k vs 80k expected.

Bad news = good news trade firing. 10-year dropping, risk ripping higher. Fed pivot narrative back in play.