Drowning in AI: Companies are launching hundreds of projects, and that’s a problem

The article argues that AI adoption is becoming universal, but too many companies are creating dozens or hundreds of low-value pilots that waste resources and reduce productivity. Executives from West Monroe, Amgen, Carvana, and GXO emphasize that AI efforts must be business-led, board- and CEO-driven, and focused on finishing a few high-impact use cases rather than spreading teams across many initiatives. The piece is commentary rather than company-specific news and is unlikely to move individual stocks meaningfully.

Analysis

The market takeaway is not “AI helps productivity,” but that implementation quality is becoming a new operating lever. That favors vendors and service providers that can prove workflow conversion, governance, and ROI measurement, while punishing enterprises that treat AI as a diffuse experimentation budget. Over the next 6-18 months, the key second-order effect is likely a widening gap between firms that consolidate AI under a single operating model and those that let dozens of internal teams duplicate spend, talent, and infrastructure.

For AMGN, this is less about drug discovery hype and more about enterprise discipline: the companies that extract value from AI fastest will be the ones that use it to compress cycle time in regulated functions without multiplying compliance risk. In healthcare, that tends to shift spending toward higher-quality tooling, validation, and data plumbing rather than open-ended pilot programs. GXO and CVNA are more exposed to execution risk because their operating models are complex and margin-sensitive; if AI creates even a modest increase in process fragmentation, the promised efficiency can slip by a quarter or two, which matters when investors are underwriting margin expansion.

The contrarian view is that the near-term “AI drag” narrative may actually be bullish for a narrower set of winners. If management teams are forced to kill weak pilots, budgets should reallocate toward a handful of enterprise platforms, systems integrators, and workflow automation names with clear payback periods. That means the current phase is likely not broad AI monetization, but a sorting mechanism: broad adoption is neutral to slightly positive for the index, while stock selection should favor names with measurable deployment economics and credible governance.

Catalyst-wise, watch for earnings commentary over the next 2 quarters that shifts from “pilot counts” to “payback period” and “workflow penetration.” The inflection point will be when companies begin quantifying labor-hours saved and headcount avoided; that’s when the market can distinguish real operating leverage from AI theater. Until then, there is a non-trivial risk that AI spend becomes a cost center disguised as innovation, especially in organizations without top-down pruning discipline.

Key Decisions for Investors

• Long MSFT vs. short a basket of lower-quality enterprise software names over 3-6 months: MSFT is better positioned to monetize AI through bundled workflow adoption, while fragmented vendors risk being displaced by consolidation. Target 1.5-2.0x risk/reward if enterprise buyers prioritize fewer platforms and lower integration overhead.
• Buy AMGN on 3-6 month horizons on pullbacks, but pair it against CVNA and GXO if you want to isolate execution quality: AMGN should benefit from disciplined, regulated AI deployment, while the logistics-heavy names face greater risk of AI pilot sprawl and margin noise. Use a 1:1 pair with tight stop on any evidence of rapid operational payback at CVNA/GXO.
• Avoid chasing broad AI-beta names for the next earnings season; instead prefer call spreads on infrastructure beneficiaries with visible conversion to revenue, not experimentation. The market is likely to reward proof-of-ROI rather than headline AI adoption, creating better skew in names with existing distribution.
• For CVNA, consider a tactical short-dated put spread into the next quarterly print if management language emphasizes experimentation over realized savings. The setup is asymmetric if investors begin to haircut margin-expansion assumptions due to execution complexity, with downside over 1-2 quarters more plausible than immediate upside.