Protect your agentic AI before you wreck your agentic AI

Senior technology executives at the Fortune Brainstorm Tech dinner at CES argued that AI should be treated as a new capability requiring fundamental redesign of processes rather than incremental automation, warning against "industrializing waste" and technology sprawl. Panelists from Deloitte, Salesforce, Disney and Solventum emphasized the need for strong leadership, clean integrated data, and engineered fail-safes (kill switches, auditability)—favoring vendors and services tied to enterprise data infrastructure, governance and AI orchestration while highlighting execution risks for companies that simply automate inefficient legacy processes.

Analysis

Market structure: AI-as-capability shifts economic rents to data-orchestration platforms (CRM, specialist healthcare stacks like SOLV) and semiconductor/cloud suppliers while penalizing firms that merely "automate existing waste" (legacy manufacturing/consulting). Expect platform pricing power to rise 5-15% over 12–24 months where integrated data + governance exist; incumbents lacking data integration will see churn and margin compression. Cross-asset: higher tech capex lifts IG corporate issuance near-term (+5–10% issuance), supports USD strength vs EM; commodity upside concentrated in semicap metals, transient for copper/palladium.

Risk assessment: Tail risks include regulatory intervention (EU AI Act national implementations, SEC/FTC auditability rules) and major operational failures (model-driven recalls or data breaches) with >10% one-day equity shocks for exposed names. Immediate (days–weeks): sentiment-driven repricing around CES/earnings; short-term (months): pilot failures and integration costs; long-term (1–3 years): winner-take-most consolidation and persistent margin reallocation. Hidden dependency: usable labeled data and orchestration tech; catalysts: large M&A (Salesforce-sized deals) or binding regulation within 60–180 days.

Trade implications: Direct plays: overweight CRM (software/platform) and SOLV (healthcare AI stack) while underweight legacy industrial software/services (select MMM exposure). Implement relative-value: long CRM vs short legacy consulting/ERP implementation names to capture faster ARR conversion. Options: buy 3-month call spreads on CRM ahead of quarterly cadence; size exposures for 1–3% portfolio conviction, target 20–30% upside in 6–12 months, stop-loss 12%.

Contrarian angles: Consensus overestimates speed of revenue conversion; integration costs and "industrialized waste" risk mean some AI hype names may underdeliver for 6–12 months—historical parallel: ERP/CRM adoption cycles had multi-year lag to margin expansion. Mispricing likely in richly valued software names without proven data ecosystems; if CRM enterprise ARR deceleration >200bps q/q, that’s a sell signal. Unintended consequence: aggressive automation can increase churn and compliance costs, creating short opportunities in high-valuation, low-data-quality names.

Key Decisions for Investors

• Establish a 2–3% portfolio long in CRM (Salesforce) within the next 2–6 weeks ahead of Q updates; target +25% in 6–12 months, set a hard stop-loss at -12% and trim if enterprise ARR growth decelerates >200bps q/q.
• Establish a 1–2% long position in SOLV as a 6–18 month thematic healthcare-AI play; add on meaningful regulatory clearance (no binding adverse guidance) and target +35–45%; reduce by 50% if a major data-privacy/regulatory action is announced within 90 days.
• Reduce exposure to legacy industrials/implementation-heavy names (example: reduce MMM-relative weight by ~50% vs benchmark over the next month) and redeploy proceeds into software (CRM) and healthcare IT (SOLV) to capture higher margin reallocation expected over 12–24 months.
• Implement options hedges: buy a 3-month CRM call spread (buy ATM, sell +20% OTM) sized 0.5–1% portfolio to limit premium; if SOLV IV <30%, buy a 3-month straddle sized 0.5% to capture event-driven move; exit positions at 30% realized move or at 90 days.
• Monitor regulatory catalysts (SEC/FTC AI auditability guidance, EU AI Act national rules) on a 30–180 day horizon; if binding audit/compliance rules are enacted, cut high-privacy-risk positions by 50% and increase allocation to vendors with certified governance stacks (CRM, proven healthcare vendors).