Jasper Built on Rented Models
Jasper
The real leverage sits below Jasper, because model APIs made it trivial for dozens of startups to launch lookalike AI apps, which pushed differentiation up into workflow, distribution, and proprietary usage data. Jasper grew fast by packaging GPT-3 into marketer friendly templates and a simple editor, but the same rented model access also let rivals and incumbents ship similar features quickly, compressing the pure app layer advantage and forcing companies like Jasper to build deeper product scaffolding around the model.
-
Jasper was effectively a reseller of foundation model output. It wrapped GPT-3 in templates for ads, blog posts, emails, and product descriptions, then charged SaaS prices on top. That helped it reach about $42.5M ARR in 2021 and roughly $75M by 2022, but it also meant every generated word carried an upstream model cost.
-
The Shopify comparison fits the supply side. OpenAI and Stability AI lowered the technical work needed to start an AI app, similar to how Shopify removed the hard parts of launching an online store. The result was a flood of text and image startups built on rented models, which made app creation easy and app defensibility weak.
-
What survived was not generic text generation, but tighter integration into business systems. Jasper moved toward a Grammarly like layer across other apps, while Copy.ai moved into structured go to market workflows inside CRM and sales tools. After ChatGPT hit free and low cost distribution in late 2022, that shift from simple writing app to embedded workflow system became necessary.
This market keeps moving toward a split where foundation models get cheaper and more interchangeable, while value accrues to apps that own the workflow, the customer context, and the feedback loop. The winners will look less like prompt boxes and more like operating systems for specific jobs, plugged into the software people already use every day.