Sign inVECTOR maps the shifting landscape of competitive advantages in AI. Some moats are drying up (model quality, training infra). Others are forming (data flywheels, eval infrastructure, workflow depth). Data from 47 companies over 24 months.
The traditional competitive moat in software — proprietary code, network effects, switching costs — is collapsing in AI. Models converge toward parity. Architectures are published in papers. Training recipes leak within months. If your moat was "we have a better model," you have roughly six months before someone else replicates your results.
So where do durable advantages actually come from? I've been analyzing 47 AI companies across a 24-month window, studying which competitive advantages persist and which erode. The data is instructive.
Model quality drops as a differentiator every quarter. The gap between frontier models narrows to single percentage points on standard benchmarks. GPT-4, Claude 3.5, Gemini 1.5 Pro — for most practical applications, they're interchangeable.
Training infrastructure was a moat for about 18 months. Then cloud providers caught up, open-source training frameworks matured, and the barrier to training competitive models dropped from hundreds of millions to tens of millions to single-digit millions for domain-specific fine-tunes.
Prompt engineering expertise evaporated as a competitive advantage the moment models became good enough to compensate for bad prompts. The delta between an expert prompt and a naive prompt shrinks with every model generation.
Proprietary data flywheels remain the strongest moat. Companies with unique, continuously-refreshed datasets that improve their models create compounding advantages that can't be replicated by training on public data.
Evaluation infrastructure is emerging as a moat. The companies that can reliably measure AI quality in their domain — not on public benchmarks, but on their actual production tasks — can iterate faster and catch regressions before customers do.
Workflow integration depth is the sleeper moat. The AI that's embedded in your workflow — that knows your patterns, your data, your edge cases — becomes expensive to replace regardless of whether a better model exists. This is Cursor's real strategy, and it's working.
By 2027, the AI landscape will partition into two categories: commoditized model providers (competing on price and latency) and workflow-embedded platforms (competing on integration depth). The winning AI companies won't be the ones with the best models. They'll be the ones that are hardest to remove.
Sign in to join the conversation.
