Are services the new software? This venture capitalist thinks the future is in selling AI-delivered outcomes, not AI-powered products

The article argues that AI-native firms could disrupt outsourced service industries by selling outcomes rather than software, with early targets including legal, accounting, insurance brokerage, IT managed services, and payroll. It highlights a potential pricing shift away from billable hours toward outcome-based billing, while noting challenges from AI inference costs and enterprise go-to-market expenses that could keep margins closer to 70% than SaaS-like 90%. Overall, the piece is a thematic industry outlook rather than a company-specific catalyst.

Analysis

The investable point is not that AI will replace software; it is that AI should compress the spread between software and labor-heavy services, but only in categories where demand is already normalized around outsourcing. That creates a winner-take-most dynamic for the first credible AI-native operators, because customers are buying a cheaper outcome, not a better tool, and procurement friction is lower once the service is already externalized. The second-order effect is that incumbents with sticky relationships may not lose volume immediately, but they will be forced to defend pricing, which should pressure EBITDA before visible revenue deterioration shows up.

The most important constraint is not model capability; it is unit economics and distribution. Inference costs are likely to stay volatile enough to cap gross margins in the near term, while enterprise go-to-market in regulated services remains slow and relationship-driven. That means the market may overestimate the speed at which pure-play AI service companies can scale like SaaS; the more likely path is months of margin compression and sales-cycle elongation before the best operators prove durable economics.

For public markets, the cleaner trade is not chasing the theme indiscriminately but separating beneficiaries of AI adoption from exposed legacy wrappers. IBM is the clearest relative loser in this set if buyers increasingly view ‘managed services’ as a commoditized outcome rather than a defensible moat; SNOW is less directly at risk but could benefit if AI service firms need data infrastructure and governance layers. UPS is mostly a neutral read-through, though the broader lesson is that any outsourced workflow business with high labor content and low differentiation faces pricing pressure from AI-native entrants. The contrarian miss is that regulatory and trust barriers may actually lengthen the transition, so the revenue displacement is more likely a 12-36 month story than a next-quarter shock.

Key Decisions for Investors

• Short IBM vs long a basket of AI infrastructure beneficiaries for 3-6 months; thesis is margin pressure in legacy managed services and consulting wrappers before demand destruction becomes visible.
• Buy SNOW on pullbacks for 6-12 months as a neutral-to-beneficiary infrastructure layer; AI-native service firms will need governance, data access, and audit trails even if they disintermediate software vendors.
• Avoid chasing pure-play AI service startups in public proxies until margin disclosures stabilize; if exposed via privates, underwrite to 70% gross margin ceilings, not SaaS-like 90%+, and demand at least 2:1 upside to downside.
• For event-driven exposure, look for any IBM rally on AI-services headlines as an opportunity to fade over 1-3 months; the first visible effect should be pricing pressure, not immediate revenue collapse.
• Pair long quality enterprise software with short labor-heavy IT/services names if available; the trade should work as customers shift from buying seats/licenses to buying outcomes over the next 4-8 quarters.