Second-order impacts + the debate:
This capex wave pulls through the stack (AI compute suppliers including $NVDA and $AMD; data center partners; power solutions). Bulls anchor to enormous RPO visibility ($523B, +438% y/y) and accelerating cloud growth; bears see “profitless growth,” legacy software pressure, and a funding puzzle—even with talk of asset-specific financing. Portrait takeaway: separate catalysts from narratives, and quality-of-earnings from headlines.
Two very different “risk resets.”
$RKLB ripped higher on technical de-risking + sector multiple expansion.
$ORCL sold off on earnings quality + a sudden step-up in capital intensity
The real shock: capex guidance jumped by ~$15B to ~$50B for FY2026.
Q2 capex printed $12.11B and FCF came in ~-$9.97B. That forces investors to re-underwrite $ORCL less like a software compounder and more like a capital-heavy infrastructure utility—where ROIC matters more than “beat and raise.”
Julian Flannery: Building a Medical Miracle Factory
It’s not everyday you get to interview one of your best friends who built the company that helped save your child’s life. 💙
Our teacher today is @JulianFlannery, Cofounder and CEO of @SummusGlobal , a company that is democratizing access to the world’s leading doctors.
We have the best collection of doctors in the world. But our healthcare system in the US is too costly, too slow and, for most, the top doctors are inaccessible.
Summus was designed to solve these existential issues with a vision to become the global hub of medical expertise for all. They have built an ecosystem of over 5,000 of the very best doctors on the planet - from hospitals like @BostonChildrens, @MDAndersonNews, @HSpecialSurgery, @nyphospital, @StanfordHealth, @UCSFHospitals and virtually all the other greats - in order to serve as the clinical front door to high quality care for millions of patients and families across ALL health questions.
As a friend of Julian’s, a board member of Summus and a father of kids who may not be here without their help, this was a very special session for me and one I’ve been wanting to have for a long time.
And if you're a decision maker in a company big or small, there may be no better thing you can do for your people or bottom line than to get on the Summus train.
Links in comments!
@portraitanalyst@PaulBuser@joincolossus@patrick_oshag
The broader takeaway: the AI trade is no longer “buy anything with GPUs.” It’s about where the bottleneck sits. $TSEM wins as SiPho capacity comes online; $CRWV stumbles as data center capacity lags. With Portrait, you can systematically track these fault lines to find the next winners and losers.
In this week’s “Portrait Weekly Winners & Losers,” I’m looking at two very different outcomes inside the AI infrastructure stack: $TSEM, a specialty foundry riding an optical supercycle, and $CRWV, a GPU cloud provider tripped up by the physical limits of data centers.
The failed Core Scientific deal was another tell. $CRWV explicitly sought more control over power and real estate, then remained dependent on third-party shells when the transaction died. Q3 simply crystallized that execution risk in the P&L and guidance.