Daily Alpha · X
· Post-Market Alpha · by Buzzberg Research
X split between macro de-rating and physical demand: the tape punished AI beta while specialists reported full backlogs, constrained capacity and selective fresh buying.
Themes on this desk
Capital competes with duration
AI corporate issuance and off-balance-sheet commitments were repeatedly linked to higher long-end yields and broader portfolio concentration.
Physical bottlenecks persist
Optics, transformers, advanced packaging and thermal equipment show demand that is limited by qualification and capacity rather than orders.
Selective dip buying
Fresh ASYS, TEM, ASTS, NBIS and SaaS additions coexisted with put hedges and explicit avoidance of weaker structures.
Ticker heat
?
Top voices by smart followers and alpha score
Market Radar →AI Corporate Debt Crowding Out Treasuries
The author argues that massive AI infrastructure debt issuance is competing with the US Treasury for capital, contributing to higher long-end interest rates.
Bond managers are selling Treasuries to buy AI corporate debt instead because it pays more.
Suggests that AI capital spending is structurally lifting borrowing costs for the broader economy.
Watch Monitor long-end Treasury yields and corporate bond issuance volumes.
Source →AAOI reports capacity constraints despite strong demand
Applied Optoelectronics CFO notes that despite interest from five potential CPO/NPO customers, the company is cautious about over-committing due to current capacity limits.
$AAOI CFO says five different customers are talking to them about CPO, with some of them also discussing about NPO.
Highlights a supply-side bottleneck for optical transceivers, potentially limiting near-term revenue growth despite strong demand.
Watch Monitor capacity expansion progress and LTA finalization.
Source →Fabrinet fundamentals show strength despite valuation compression
Fabrinet reported Q1 revenue of $1.316B (+45% YoY) and adjusted EPS of $4.10, with management citing 'insatiable' demand for DCI/transceivers and a capacity roadmap supporting $12.5B-$14B in revenue.
$FN earnings TLDR: > Revenue $1.316B, +45% YoY, above guide > Adj EPS $4.10 > Q1 guide $1.375B-$1.425B, ~43% YoY
The stock experienced a 21% price decline while revenue and EPS run-rates increased by 10%, suggesting a potential valuation disconnect.
Watch Monitor if the 800G hyperscaler-direct ramps and merchant programs starting in late 2026/early 2027 meet the accelerated growth targets.
Source →Global LLM model-layer revenue run-rate projected at $423B by December 2027
The global LLM model-layer annualized revenue run-rate is estimated at $422.8B for December 2027, with a bear case of $254.3B and a bull case of $703.5B. Current run-rate is estimated at $146.4B.
EXECUTIVE CONCLUSION The central estimate for global LLM model-layer annualized revenue run-rate at December 2027 is $422.8B, with a bear
Provides a framework for evaluating the scale of the AI model-provider ecosystem, distinguishing between pure API revenue and broader provider-owned monetization.
Watch Monitor if industry spending growth persistently trails token growth by >25-30 percentage points, signaling accelerating commoditization.
Source →ASYS long thesis on AI packaging
The author is long ASYS, citing strong AI-related revenue growth (~120% YoY) in its thermal processing segment and high margins as evidence of a compelling growth story.
I'm now long $ASYS, Nice little dip. > Babyfolio (@babyfolio) > I think $ASYS is a very interesting company.
Highlights a small-cap play on the increasing complexity of AI chip packaging.
Watch Monitor backlog growth and execution during the transition to a semi-fabless model.
Source →