Is AI fueling a new business model? A glimpse into former OpenAI research director Bob McGrew's AI investment framework

Former OpenAI Chief Research Officer and core leader at Palantir, Bob McGrew, shares his perspective on the current AI boom from an investor’s point of view. He candidly states that the market atmosphere is very much like a bubble period, with billions of dollars being raised everywhere and every pitch deck must include AI-related investments. However, in such an environment, McGrew does not invest in every project but follows a clear AI investment framework.

The AI Boom Arrives, Market Atmosphere Resembles a Bubble

At the start of the interview, the host describes the current AI investment climate as reminiscent of past bubble eras. Constantly, there are fundraising rounds of one hundred million or two hundred million dollars, and nearly every team presenting has a dedicated segment discussing AI.

The host also shares his own experience. He previously managed a legal tech SaaS company that, after collaborating with OpenAI and integrating AI to assist legal assistants with their work, saw a significant increase in value. It was eventually acquired by Thomson Reuters at a valuation eight times higher than the last funding round, demonstrating that AI is indeed rapidly transforming many industries.

Against this backdrop, the host asks McGrew, besides OpenAI, which AI startup directions in the current market are truly worth investing in, and whether infrastructure remains a necessary area for strategic layout.

(Note: Thomson Reuters is a global professional information and data services company, mainly serving legal, financial, accounting, tax, risk management, and media professionals.)

Maintaining Distance from AI Infrastructure, Focusing on Unprecedented Applications

McGrew openly admits that he remains relatively reserved about investing in startups solely focused on AI infrastructure. His reasoning is that most infrastructure addresses issues arising from “current models,” but with the advent of LLMs like GPT-5, usage patterns and demand structures are likely to undergo fundamental changes. Infrastructure built now may not be suitable in the future.

Instead, he prefers to see applications that use AI to solve “problems that were impossible to solve before,” rather than simply adding AI to existing processes to create an “AI version.”

He also clearly states that his greatest interest lies in innovative teams at the application layer.

Unlimited Interns and New Business Models Expected to Emerge

Regarding the value of application-layer innovations, McGrew uses a very concrete analogy to explain his investment logic. He describes current AI as suddenly having “an unlimited number of short-attention-span interns.”

In the past, many things couldn’t be done not because of technical limitations but because of high labor costs, slow speeds, and management overhead. But if a large group of “intern-level” labor suddenly appears, many previously unreasonable or unprofitable tasks become feasible.

He also compares model capabilities to students: GPT-3 is roughly like a high school student, GPT-3.5 like a freshman, GPT-4 like a junior, and GPT-5 will be at a completely new level. In his investment judgment, the key is never how strong the model itself is, but whether entrepreneurs have thought through what new workflows and business models could emerge if they truly had this “infinite number of interns.”

Avoiding Direct Confrontation with Hardware and Computing Giants, While Betting on New AI Technologies

Regarding the influx of capital into GPUs, data centers, and computing power leasing, McGrew’s view is pragmatic. He points out that many have attempted to develop new chips in the past, but currently, NVIDIA’s dominant market and capital advantages allow it to continuously improve its products. Therefore, betting directly on NVIDIA to win the compute war is very challenging.

However, he does not completely dismiss such investments. Instead, he believes that maintaining a small portion of the investment portfolio to bet on “high-risk, high-reward” new technologies that could become huge if successful remains a reasonable strategy.

The Core of the Investment Framework: What Will AI Rewrite and Make Worth Doing

Finally, McGrew shares his AI investment framework. The core is not about chasing the hottest technical buzzwords but about thinking through how AI will change “what is worth doing.”

He states that truly valuable entrepreneurial opportunities are not about making existing processes slightly faster but about enabling things that were previously impossible due to manpower, cost, or scale issues—making them feasible for the first time, or even creating entirely new industries.

(Future Business Models of AI Remain Unpredictable, a16z Analyzes the Next Wave of AI Technology and Investment Changes)

This article, “Will AI Ignite New Business Models? A Look into Former OpenAI Research Director Bob McGrew’s AI Investment Framework,” first appeared on Chain News ABMedia.