---
id: "claim-custom-models-outsourced"
type: "claim"
source_timestamps: ["§ Could You Build a Better Program?"]
tags: ["build-vs-buy", "foundational-models"]
related: ["entity-openai", "entity-midjourney", "action-outsource-general-ai"]
confidence: "high"
testable: true
speakers: ["Jay B. Barney", "Martin Reeves"]
sources: ["spine"]
sourceVaultSlug: "hbr-seg-spine"
originDay: 1
articleStem: "hbr-cl-96-ai-no-sustainable-advantage"
sourceUrl: "https://hbr.org/2024/09/ai-wont-give-you-a-new-sustainable-advantage"
sourceTitle: "AI Won’t Give You a New Sustainable Advantage"
---
# Building custom foundational models is inferior to outsourcing to general-purpose providers

**Claim (confidence: high, testable):** It is highly unlikely that a user company has the resources to build a *better* general-purpose Gen AI platform than specialized providers such as [[entity-openai-d1]] or [[entity-midjourney]], who hold years of scaling experience. Even if a special-purpose AI is built, competitors will quickly develop, cooperate to build, or outsource their own equivalents — making the advantage temporary. The strategic implication is [[action-outsource-general-ai]].

**Enrichment / nuance:** Supported for *frontier, general-purpose* models: training them requires massive compute, data, and specialized expertise, effectively limiting it to a small set of AI labs and hyperscalers — most firms are better off buying/partnering. **But 'inferior' is too broad if read to cover all in-house development:** the competitive-advantage literature notes that *domain-specific* models — fine-tuned or purpose-built on proprietary data and embedded deeply into workflows — can be defensible even when the base model is commoditized. Valid in the general-purpose sense; qualified for narrow, domain-specific systems.