---
id: "concept-stage-gates"
type: "concept"
source_timestamps: ["§ Why Take a Step-by-Step Approach to AI Portfolio Management?", "§ How the Portfolio Runs"]
source_url: "https://hbr.org/2026/01/manage-your-ai-investments-like-a-portfolio"
source_title: "Manage Your AI Investments Like a Portfolio"
tags: ["governance", "risk-management", "project-progression"]
related: ["concept-dual-lens-portfolio", "framework-four-portfolio-stages"]
definition: "Tightly defined go/no-go progression tests applied at each transition in the AI portfolio pipeline to assess feasibility, strategic fit, and readiness."
sources: ["spine"]
sourceVaultSlug: "hbr-seg-spine"
originDay: 1
articleStem: "hbr-foci-61-ai-investments-portfolio"
sourceUrl: "https://hbr.org/2026/01/manage-your-ai-investments-like-a-portfolio"
sourceTitle: "Manage Your AI Investments Like a Portfolio"
---
# AI Stage Gates

> **Definition:** Tightly defined go/no-go progression tests applied at each transition in the AI portfolio pipeline to assess feasibility, strategic fit, and readiness.

Borrowed from traditional R&D and new-product development, stage gates are tightly defined go/no-go progression tests that protect organizations from a flood of strategically disconnected AI pilots (the failure mode described in [[claim-piecemeal-drain]]).

At each portfolio transition, a candidate project is judged on three things: its standalone merits, its relation to competing opportunities, and cross-initiative dependencies. The gates ask critical questions:

- Is the required **data** available, high-quality, and governed?
- Are the necessary **skills** sourced?
- Does the supported **business process** require redesign?
- Are **ethical and security controls** established?
- Does the **business case** remain valid?

By enforcing these gates, companies ensure only worthy, viable AI innovations reach production, while struggling or misaligned efforts are redirected or retired. Stage gates are the second of the [[framework-three-portfolio-mechanisms]] and structure every transition in the [[framework-four-portfolio-stages]]. They presuppose [[prereq-stage-gate-processes]]. Their calibration for less-regulated industries is an open issue — see [[question-low-regulation-adaptation]].

**External grounding:** This is a direct adaptation of R. G. Cooper's classic **Stage-Gate** model for innovation governance.

**Counter-perspective:** Heavy gating can slow exploratory work in fast-moving AI domains; some digital-native firms minimize formal gates to preserve speed, favoring tiered/federated governance.