OpenAI targets chip design and enterprise sectors, arguing efficiency beats open-source costs

OpenAI is expanding its advanced models into sectors such as chip design, life sciences, and finance. [3] The company argues that a model with a higher upfront cost can ultimately be more cost-effective if it successfully performs a task with fewer tries. [3] Instead of evaluating models based on token price, OpenAI wants customers to consider which model provides the cheapest completed task. [3] At a Goldman Sachs conference in San Francisco, OpenAI CFO Sarah Friar noted growing demand for AI developed for specific tasks. [3]

In the chip design sector, OpenAI is not trying to replace existing electronic design automation systems. [3] Instead, its solutions assist engineers by helping them analyze problems, evaluate approaches, and accelerate certain tasks. [3] To demonstrate this, OpenAI cited its first custom inference chip, Jalapeño. [3] The company used its AI technology to move Jalapeño from conception to tapeout in just nine months. [3] AI shortened the cycles of design, measurement, and verification, and was also used to improve arithmetic circuits and program the final chip. [3]

OpenAI is also experimenting with a new business model where pricing is based on the AI's effectiveness for a given company, rather than purely on consumption. [3]

OpenAI is attempting to reframe the enterprise AI value proposition from raw computing cost to workflow efficiency, a shift necessary to justify its premium pricing against increasingly capable open-source alternatives. By highlighting its internal success with the Jalapeño chip, the company offers a concrete proof point for complex, high-value tasks. The move toward outcome-based pricing could appeal to industries like finance and life sciences, where the cost of errors is high. What the sources don't address: How OpenAI plans to objectively measure and verify "effectiveness" to determine billing under an outcome-based pricing model.