The S&P 500 is flat, but capital is rotating into AI infrastructure choke points as 2026 AI CapEx steps up ~70% YoY.
NVIDIA’s CPO shift confirms value is moving deeper into optics, memory, and semicap tools.
This is a positioning call on suppliers with pricing power and volume leverage. We break down the tickers best set up to rerate.
Key Report Insights:
Global Rotation: $65B/week flows ex-U.S.; EM AI rerates at ~10x forward P/E
Optics Inflection: NVIDIA embeds CPO; lasers move inside the chip stack
Edge Expansion: FPGA attach rates double; ASPs rise with server content
CapEx Flywheel: AI spend +73% YoY; WFE pricing power tightens
Process Intensity: HBM + advanced packaging drive >20% semis growth guide
1. Market Overview:
A. The Emerging Premium:
Global equity allocations to U.S. stocks have fallen to 26%, down sharply from their 92% peak in 2022.
Capital is rotating internationally, with global funds absorbing ~$65B per week into ex-U.S. equities.
Emerging markets are key beneficiaries — offering AI-linked exposure at materially lower valuations than U.S. peers.
South Korea is a case in point: up +159% YoY on memory momentum, yet still trading at ~10x forward earnings.
Source: The Kobeissi Letter on X
B. Concentration Is No More:
The S&P 500 is trapped in a 3.7% two-month range — one of the tightest compressions on record.
Beneath the surface, dispersion is extreme: single stocks are realizing ~25 volatility points above the index, near historic highs.
Leadership is bifurcated. AI hardware — tools, power, infrastructure — drives upside. AI disruption fears pressure software on the downside.
Source: Daily Chartbook on X
2. Lattice Semiconductors (LSCC):
A. AI Goes On Edge:
LLMs today are cloud-dependent, anchored to centralized data centers. But distilled models now deliver good-enough performance on device.
As AI expands into agents, humanoids, and FSD — where latency and privacy matter — the world shifts to a hybrid cloud-edge stack.
Source: CSEM
B. Lattice - The Bedrock of Edge AI:
We added Lattice Semiconductor (LSCC) to our mock portfolio on Oct 23, 2025 — it’s up +41.1% since.
LSCC specializes in FPGAs — programmable chips that create custom data paths tailored to specific workloads.
These are critical for edge AI, where compute is constrained and efficient, low-latency data flow matters.
With exposure spanning GPUs, vision systems, robotics, and wearables — and a +60%FPGA portfolio expansion in 2025 — LSCC sits squarely at the center of edge adoption.
Source: Lattice Semiconductors
This post is for paid subscribers
 | A guest post by
|