Isn't that Special!
The NY Knicks Are Special!
It might be sacrilegious to even be thinking about stocks or the markets today, with the New York Knicks having just won their first championship since 1973. But duty calls—and the specialists gave us some especially juicy quotes to share this week.
THE SECTOR SPECIALIST WEEK IN QUOTES
June 8–12, 2026
Rank 1 | Mark Schilsky — J.P. Morgan — June 9
The two datapoints that could finally break the AI trade
Quote: “Two are worth highlighting: 1) A slowdown in the growth of the annualized run-rate revenues of the major AI labs. If there is any sort of second derivative ‘kink’ in their growth algorithms, that could portend a future problem for the AI trade. 2) Scaling laws being upheld. We really haven’t had fully Blackwell-trained models on the market quite yet. No one truly knows if Blackwell will result in scaling laws being upheld…They better be great…”
Why it ranks: Our sentiments exactly! Look for the next ARR updates and their trajectory. My own guess is Theinformation.com will break the story or Dylan Patel. AI stocks won’t crack because of valuation alone; it cracks if lab revenue decelerates or the next generation of models fails to show a meaningful capability jump.
Rank 2 | Blayne Curtis — Jefferies — June 11
WFE estimates may still be dramatically too low
Quote: “WFE spend remains strong with clean room availability gating some shipments this year, suggesting C27 and even C28 could see even stronger growth as those fabs are brought online. There is likely upside to our WFE forecast of $148/181/202B for C26/27/28 with C27 likely trending to $200B+ and C28 potentially $250B+. The higher spend is levered to the AI build out, and demand data points remain extremely positive despite rising concerns of a potential pullback in Enterprise AI spend.”
Rank 3 | Duncan Wagner — J.P. Morgan — June 8
SOCAMM content reduction is a supply problem, not an AI-demand problem
Quote: “NVDA and Memory makers are introducing a 96GB SOCAMM version, half of initial plan 192GB (total module 768GB vs 1.5TB) due to supply constraints as SOCAMM shares capacity w/ Mobile DRAM. If NVDA maintained their original 192GB plan, supply would be insufficient given robust Vera CPU demand. Investors generally concur supply remains the culprit vs some market claims of excess power consumption…Constraints continue to bode positive for HBM/DRAM capex spend.”
Rank 4 | Joshua Meyers — J.P. Morgan — June 11
Optical selloff and CPO-delay fears
Quote: “Samik Chatterjee pushes back on the recent ~20% pullback in optical names…arguing that CPO delay fears are overblown and the setup is increasingly attractive at current valuations. The key data point: NVDA’s 2H26 CPO ramp is tracking better than expected, with LITE flagging CPO-supported switch mix now trending >50% (vs. 40-50% prior), and Nvidia actively seeking additional InP capacity from LITE — potentially prompting a sixth fab even before the current one ramps.”
Why it ranks: Quite the panic but late in the week we saw that Dylan & SemiAnalysis stand by their call fwiw.
Rank 5 | Andrew Rocha — Wells Fargo — June 9
Token optimization versus structural AI-demand destruction
Quote: “Tokenmaxxing is likely temporary and he expects it to flip to outputmaxxing or tokens-at-a-reasonable-cost. Fluidstack believes token consumption is more likely to follow value and output over time. Fluidstack currently spends 7-figures on token consumption annually.”
Why it ranks: Oh Andrew. We aren’t sure Fluidstack is the best resource on this topic. The list of token max companies protesting is growing larger. Our take is perhaps its optimization, a quick real dip in ARR that then grows at a rate that isn’t quite as steep? The more important data point might even be we need to now see the profit models of these companies at $50b ARR. Isn’t that a large enough base to show its a good business? Look for the S-1s to move from confidential to available for reading.
