AI Barbell Strategy: Long Infrastructure, Short Disrupted Incumbents
The AI Barbell is Closelook's core portfolio construction principle: concentrate the long side in AI infrastructure companies that benefit regardless of which applications win (Natural Position in ABR terms), and the short side in companies whose core business model autonomous agents structurally replace (Terminal in ABR terms). The barbell shape means minimal exposure to the messy middle — the Cannibalize or Die companies where outcomes are uncertain and execution-dependent. This is not a market-neutral strategy; it has a structural long bias because infrastructure build-out is the higher-certainty trend.
The Long Side: Infrastructure
Infrastructure providers benefit from the AI build-out regardless of which AI applications, models, or agents win. TSMC fabricates chips for NVIDIA, AMD, Google, and Amazon. ASML's EUV machines are required by every advanced fab. Broadcom's networking chips connect every AI cluster. These companies have structural demand driven by physics, not hype.
The Functional Index provides the long-side universe. The 6 layers ensure diversification across the supply chain rather than concentration in Compute (Layer 4) alone.
The Short Side: Disrupted SaaS
The short side targets companies scored as Terminal by the ABR Framework. These are companies whose core value proposition — the thing customers pay for — is something an autonomous agent replicates at near-zero marginal cost. The SaaSpocalypse thesis provides the structural argument.
Why Skip the Middle
Cannibalize or Die companies (Microsoft, Salesforce, Adobe) are the hardest to position on. They have the resources and distribution to pivot, but pivoting means cannibalizing their own revenue. The outcome is binary and execution-dependent. The AI Barbell deliberately avoids these uncertain positions in favor of higher-conviction long and short legs.