Tech Disruptors: Domino Data Lab CEO on Taking AI to Production

AI's enterprise expansion is exposing a gap between rapid prototyping and reliable, organization-wide deployment, creating a need for governance and standardization. Domino Data Lab CEO Nick Elprin says model-driven organizations are putting AI into core-business settings and that coding assistants enabling domain experts to build full-stack analytics increase demand for operational platforms and governance.

Analysis

The real battleground is not model accuracy but operational control: organizations will prioritize auditability, lineage, and rollback over marginal model improvements. Expect IT budgets to shift meaningfully toward observability/MLOps tooling — in our view 10–20% of incremental AI spend over the next 12–24 months will be allocated to governance and deployment plumbing, creating a multi-year revenue tail for platforms that own the data/control plane rather than point-model vendors.

A predictable second-order outcome is consolidation around vendors embedded in the data layer and IT workflows (cloud providers, data warehouses, ITSM). Coding assistants will democratize app-building, but that increases shadow deployments and compliance incidents, which in turn accelerates demand for security/logging and policy automation; vendors that offer native policy hooks will capture disproportionate margin. Conversely, high-margin bespoke model integration services are at risk as enterprises standardize on repeatable stacks and internalize domain experts with low-code assistants.

Tail risks are event-driven: a high-profile model failure, data exfiltration, or rapid regulatory mandates (privacy or model explainability) could compress valuations for “pure-play” model vendors within weeks and redirect budgets back to incumbents with compliance pedigrees. A countervailing catalyst that would reverse consolidation is rapid commoditization of LLMs combined with open-source toolchains that reduce lock-in — that would favor specialist feature vendors and push platform economics outward rather than inward over 12–36 months.

Key Decisions for Investors

• Pair trade (12–18 months): Long SNOW (Snowflake) vs Short PLTR (Palantir). Rationale: SNOW benefits from capturing the data/control plane and standardized governance hooks; PLTR is priced for bespoke deployments which should compress. Target: SNOW +40% / PLTR -30%; stop-loss: -20% on SNOW leg.
• Long CRWD (CrowdStrike) via 12-month call spread to fund premium (~6–12 months). Rationale: rising demand for telemetry and model-behavior monitoring after shadow deployments; cyber vendors can re-price recurring telemetry. Risk/Reward: aim for 2:1 upside (30–50% return) vs max premium loss (~100% of option premium).
• Long NOW (ServiceNow) 9–15 months: buy calls or outright stock. Rationale: ITSM and approval workflows become binding controls over domain-expert-built apps; NOW can upsell governance modules. Target +30% with a tactical stop at -15%.
• Short AI (C3.ai) 6–12 months: outright short or buy put spread. Rationale: pure-play AI software without deep data-plane integration is exposed to margin contraction as customers consolidate; high multiples re-rate on missed enterprise governance wins. Reward: asymmetric if consolidation accelerates; risk: platform partnerships could buoy revenue — cap position size to 1–2% NAV.