Former OpenAI CTO Mira Murati is reportedly preparing a new move in artificial intelligence: an independent startup focused on AI products built with proprietary models. The talks are still early, but the scale of the possible fundraising has already made the effort notable.
According to sources speaking to Reuters, Murati is in discussions with venture capitalists for the new company. Sources say the startup could potentially raise over $100 million, although exact funding targets have not been disclosed.
What Murati is reportedly building
The planned company would develop AI products based on proprietary models. That detail matters because building custom AI models is capital-intensive, and it signals a more ambitious path than simply packaging existing systems into applications.
The source article does not say what products the company would offer, what markets it would target, or when it could launch. It also remains unclear whether Murati would serve as CEO.
What is clear is that Murati’s background gives the reported venture immediate attention. At OpenAI, she played a key role in projects including ChatGPT and DALL-E, and she was also involved in the company’s partnership with Microsoft.
The new company may also receive support from Barret Zoph, OpenAI’s former head of post-training for AI models. Zoph left OpenAI on the same day as Murati in late September.
Why proprietary models are a difficult bet
A startup built around proprietary AI models faces a demanding market. The source describes the AI model market as a red ocean, with many commercial and open-source models offering similar performance levels.
That creates a hard strategic problem for any new entrant. If models quickly look similar from the outside, a new company has to show why its approach is meaningfully different, whether through product design, model capability, deployment strategy, or another advantage already implicit in the founding team’s vision.
The cost side is also central. The source notes that AI models can cost millions of dollars to develop and may become outdated within weeks. That combination helps explain why a funding round above $100 million could be discussed, even at an early stage.
For Murati’s reported startup, the key question is not only whether venture capitalists will back the company. It is whether proprietary model development can create enough differentiation in a crowded and fast-moving field.
Part of a broader OpenAI alumni shift
Murati is not the only former OpenAI leader to pursue a new direction. Several prominent former staff members have recently left OpenAI for startups, competitors, or new projects.
Ilya Sutskever, co-founder and former chief scientist of OpenAI, has launched Safe Superintelligence Inc., a company focused on developing safe ASI. That startup recently raised $1 billion in funding, despite being described in the source as essentially a research project with no short- or medium-term revenue in sight.
Andrej Karpathy, a leading AI researcher at OpenAI and former head of AI at Tesla, founded a company aimed at creating new AI-based educational opportunities.
Other departures have moved talent directly into rival AI organizations. Jan Leike, former co-leader of OpenAI’s Superalignment team, joined Anthropic to work on AI safety. John Schulman, another OpenAI co-founder and a key developer of ChatGPT, also joined Anthropic.
The reasons behind these exits vary. The source points to concerns about OpenAI’s priorities and internal conflicts among the factors involved. Taken together, the moves suggest that former OpenAI employees still see major potential in AI, but not necessarily through one company or one strategy.
What the departures suggest about AI competition
The reported Murati startup adds another data point to a larger question: how durable is OpenAI’s lead? The source frames the wave of departures and new ventures as a reason to reconsider what well-informed former employees believe about the field’s future.
If a company were close to releasing transformative artificial general intelligence, described in the source as ASI or AGI, it would raise a natural question about why former employees would launch competing ventures. It would also raise a question about why investors would fund those efforts.
That does not prove any single conclusion about OpenAI, Murati, or the direction of AI research. But it does show that experienced AI leaders are still treating the market as open enough to justify new companies, new research agendas, and new products.
For Murati, the reported plan appears to be a direct test of that opening. A startup centered on proprietary models would enter one of the most competitive areas in technology, but it would also be led by someone closely associated with some of OpenAI’s best-known work.
The next important details are still unknown: the company’s leadership structure, funding target, product focus, and technical direction. Until those emerge, the report mainly shows that another major OpenAI alumnus may be preparing to turn experience inside a leading AI lab into a new independent venture.