TheValueist @TheValueist Nov 8, 2025 $NVDA $GEV $VRT $CIEN The Citrini Research report documents at-site evidence tha...
Rotate from a ‘semis-only AI’ framing to an ‘AI industrial capex’ framing. The Citrini Research notes on-site evidence that large AI campus buildouts are being driven by power, cooling, and networking needs as much as by compute. Positioning should overweight infrastructure beneficiaries while maintaining exposure to compute.
Linked assets
This thesis highlights four tickers: $VRT and $GEV as critical power and thermal infrastructure plays, $CIEN for optical and networking interconnects, and $NVDA as the compute anchor that remains necessary but may not capture the full cycle’s upside.
Vertiv Holdings Co. provides critical power and thermal infrastructure solutions for data centers.
Critical power and thermal infrastructure are core to ‘power/infrastructure-led’ data center builds.
Global Energy Ventures (GEV) — positioned as a provider relevant to grid and power equipment scaling for large campuses.
Power- and infrastructure-led framing directly maps to grid/power equipment demand as campuses scale.
Ciena Corporation supplies optical networking equipment used for campus interconnects and large-scale data center networks.
Optical/network buildouts tend to scale with campus size and interconnect requirements.
NVIDIA Corporation operates as a data center scale AI infrastructure company.
Compute remains required for AI data centers, but the post suggests broader bottlenecks; benefit may be less singular vs infra names.
Source proof
Source proof: Strong source proof | 6 extracted claims | 4 directional assets | 1 supporting author | 4 successful tracked legs | headline-like title review
The underlying posts cite a Citrini Research on-site report describing multi-building AI campus buildouts (e.g., Abilene “Stargate”) and argue the cycle is power- and infrastructure-led rather than purely semiconductor-driven. Related posts discuss cloud/accelerator access dynamics and institutional data (e.g., Bloomberg/Goldman terminal notes).
Post is informational about Bloomberg Terminal access to Goldman Sachs ($GS) equity baskets; no market thesis, catalyst, positioning, or trade implication beyond mentioning $GS entitlements and “GIR Portfolio Strategy — Shareholder Return / Cash” basket label.
Post argues that investors are mis-modeling Meta’s ROIC and the financial impact of a reported “conversion lift.” In a target-CPA/ROAS ad auction, incremental conversion performance tends to be competed away/"capitalized" into auction dynamics rather than translating one-for-one into revenue or margin expansion. Text is truncated, limiting specificity and catalyst linkage.
Post cites a Citrini Research report with on-site evidence that the AI data center cycle is primarily a power- and infrastructure-led industrial investment wave (not just a semiconductor upcycle). Mentions Abilene “Stargate” complex described as 8 buildings, implying large-scale buildout. Cashtags: $NVDA $GEV $VRT $CIEN.
Source claims, based on alleged conversations with Google infrastructure personnel, that Alphabet is far from (or may never) selling TPUs directly to third parties; TPU access would remain via leasing through Google Cloud (GCP) because large-scale TPU deployment is difficult and Google will not disclose proprietary IP. Actionable mainly as a competitive-positioning datapoint for AI accelerator market structure (merchant GPUs vs closed TPU ecosystem) and for GOOGL cloud moat narrative.
Post is a pinned talk/presentation teaser describing a discretionary long/short equity framework focused on “counterparty analysis” (who’s on the other side of the trade) vs. abstract valuation. No explicit tickers, catalysts, positions, or tradeable claims are provided in the excerpt.
Supporting authors
Primary author: TheValueist (@TheValueist). Related commentary aggregates Citrini Research field observations and TheValueist’s interpretation of infrastructure vs. compute drivers.
Unlock full thesis monitoring
Consider reweighting portfolios toward power, cooling, and networking suppliers alongside existing compute exposure; review each company’s exposure to large-scale data center campus projects and customer mix before sizing positions.