Sam Altman: GPT-4 Did Not Disrupt Business as Fast as He Predicted

In a podcast interview with David Senra, OpenAI CEO Sam Altman admitted he misjudged how quickly AI would reshape businesses after GPT-4 arrived in 2023. He expected software companies to face rapid disruption but says economic inertia slowed things down. He also disclosed that OpenAI was significantly behind Anthropic's Claude Code before deciding to build its own competing coding tool, Codex, a move he described as a near-impossible bet at the time.

What happened

Detail	Fact
Speaker	Sam Altman, CEO of OpenAI
Venue	Podcast interview with David Senra
Model referenced	GPT-4, released 2023
Competing product	Anthropic’s Claude Code
OpenAI’s response	Codex coding product

Sam Altman told podcaster David Senra that he expected a wave of software business disruption to follow GPT-4’s 2023 launch. In his words: “I thought when we got to GPT-4, which was back in 2023, that very quickly after that there was going to be much more disruption, software businesses up for grabs right away, than it turned out to be.”

He pointed to one main cause: the economy resists sudden change. “The economy just has so much inertia,” he said. In other words, even when the technology is ready, the organizations, workflows, and incentives that surround it take much longer to shift.

Why it matters

Altman’s admission puts a useful frame around the current AI moment. Model capabilities have advanced quickly. Real-world adoption has not kept pace. For businesses evaluating AI investment, that gap has a practical meaning: the returns are likely to come later and more gradually than the headlines suggest.

At the same time, the gap will close. The question for any business is whether it is building the internal readiness now, or waiting until adoption is already mainstream and the competitive advantage is gone. You can read more about how enterprises are already navigating this shift in our coverage of where AI model deployment is heading next.

The Claude Code story: a competitor already had momentum

Altman also spoke candidly about OpenAI’s decision to enter the AI coding assistant market with Codex. At the time, Anthropic’s Claude Code had already built a strong user base and significant momentum.

Altman’s description of the situation was blunt: “We were way behind Claude Code, and it seemed like a kind of crazy kamikaze mission to try to beat them with a coding app.” He added that conventional business logic would say to walk away: “The consensus is that this kind of thing never works, and you just move on to the next one. It’s a fool’s errand to try to win when someone else already has momentum in a particular product category.”

Despite those odds, Altman says Codex has since become a commonly used tool among the top programmers he knows personally. He did not provide user numbers or market share figures.

Our take

Altman is right about inertia, and it is worth taking seriously coming from someone with a direct view into enterprise adoption. The businesses we work with are not slow because they are skeptical of AI. They are slow because changing a workflow means retraining people, updating integrations, and accepting short-term disruption for long-term gain. That friction is real and it does not disappear just because a model gets smarter.

The Codex story is also worth noting. When a well-funded company with a strong brand says it was “way behind” a competitor and chose to compete anyway, and wins, it signals that first-mover advantage in AI products is more fragile than in traditional software. Distribution, trust, and iteration speed can close a gap faster than usual. That cuts both ways for any business building on top of AI tools: your current vendor’s lead is not guaranteed.

If you are exploring how to weave AI tools into your own business processes, our AI integration service covers practical implementation rather than theoretical roadmaps. And if runaway AI costs are a concern, the story of an AI agent that spent $1,000 while its CEO was at dinner is a good reminder that governance matters as much as capability.

What to do about it

1. Audit which AI tools your team is actually using versus which ones were trialed and abandoned. The gap tells you where inertia lives in your organization.
2. Pick one workflow to automate or augment with AI in the next 30 days. Small, finished wins build organizational confidence faster than big pilots.
3. Track competitor AI tool adoption, not just capability announcements. Claude Code vs. Codex shows that momentum shifts can happen faster than expected.
4. Set a review date six months out to reassess. Altman’s own miscalculation shows that even insiders misjudge the pace. Build in checkpoints rather than betting on a single timeline.

The practical takeaway: plan for AI to matter more to your business in three years than it does today, but do not freeze your roadmap waiting for a disruption wave that may arrive gradually rather than all at once.

Source: Bing News · OpenAI