McKinsey AI Leader on the State of AI Development

McKinsey's Kate Smaje said the debate over potentially slowing AI development and AGI risks contrasts with boardroom priorities: companies that have already invested heavily in AI are focused on scaling deployments, generating profits, and ensuring sustainability. The discussion highlights uncertainty around the pace of AI progress and the commercial returns on enterprise AI spending, without citing specific financial results or forecasts.

Analysis

The investable issue is shifting from model capability to enterprise conversion: whether pilot deployments become recurring, workflow-embedded spend rather than discretionary experimentation. That favors vendors with distribution, proprietary enterprise data and measurable labor/productivity ROI—MSFT, ORCL, NOW, CRM and SAP—over model-adjacent names whose valuations assume sustained frontier-model capex regardless of customer monetization. Over the next 1-3 quarters, AI-related guidance will increasingly be judged on incremental revenue, retention and gross-margin contribution rather than contract announcements or seat-count adoption.

A second-order implication is a potential capex digestion cycle. If CIOs lengthen deployment timelines while assessing governance, integration and unit economics, hyperscaler GPU and data-center spending can remain elevated but become more uneven; suppliers with the highest expectations and least diversified demand are most exposed to estimate resets. NVDA remains structurally advantaged, but the market’s key sensitivity is not AI demand direction—it is whether cloud customers can show AI revenue scaling fast enough to sustain current infrastructure intensity without worsening FCF conversion.

Contrarian view: cautious boardroom rhetoric need not mean an AI spending bust. It may concentrate budgets in a smaller set of production-grade platforms, raising switching costs and expanding share for incumbent enterprise software. The decisive evidence is FY26 guidance: accelerating AI annual recurring revenue, services attach rates and stable gross margins would validate monetization; rising implementation costs, delayed go-lives, or a widening gap between capex and AI revenue would falsify the bullish platform thesis within 6-12 months.

Key Decisions for Investors

• Favor a 6-12 month quality pair: long MSFT or ORCL / short a high-multiple, less-diversified AI infrastructure basket via SOXX. Thesis is enterprise monetization accrues first to distribution and installed-base owners; reassess if hyperscaler AI revenue growth materially outpaces software AI bookings for two consecutive quarters.
• Add NOW on post-earnings weakness rather than chase broad AI beta: its workflow ownership provides a clearer path from automation to labor-cost ROI. Target a 12-month horizon; exit on decelerating subscription cRPO plus evidence that AI attach requires discounting or compresses gross margin.
• Treat NVDA as a catalyst watch, not a fresh directional recommendation without cloud-provider capex and AI-revenue disclosure. A 1-3 month downside hedge through put spreads is warranted if hyperscaler guidance shows capex persistence but no corresponding AI revenue/KPI improvement; the core risk is continued supply tightness and upside revisions.
• Monitor CRM and SAP for AI booking disclosure, implementation duration and renewal uplift at the next results. Initiate only if management quantifies production deployments and contribution to remaining performance obligations; generic usage metrics alone do not support further multiple expansion.