Unbearable, Ilya fights Ultraman in the palace. Microsoft CTO reveals the inside story: Is it all because he is jealous that his subordinates are too good?

In 2023, OpenAI staged a Silicon Valley version of "Game of Thrones": Ultraman was kicked out of OpenAI by the then chief technology officer Mira Murati and chief scientist Ilya without knowing anything about it. But with such horse-trading drama, do you still think it is a noble showdown of AI safety concepts? No, this time, the email stated the truth in a dazzling light: insufficient computing power, tight manpower, and product explosions squeezing the research budget...

Sam Altman kicked out of OpenAI

In addition, Jakub Pachocki used "amazing progress" to overcome the problem that his mentor Ilya had been stuck in for many years. Geniuses get injured and organizations get out of control.

As a result, one of the most outrageous "palace fights" in the technology circle was ignited in the flames of jealousy.

Master and apprentice fight, Ultraman is expelled

An email from Microsoft Chief Technology Officer Kevin Scott completely revealed the truth behind Ilya’s departure from OpenAI─

Ilya did not hesitate to detonate the OpenAI crisis due to a deep and unspeakable "intellectual jealousy" of Jakub Pachocki.

Ilya Sutskever (left) and Jakub Pachocki (right)

The latest legal documents disclosed in the Musk v. Altman case in 2026 have rarely revealed the personnel undercurrents behind the OpenAI crisis.

In November 2023, Microsoft Chief Technology Officer Kevin Scott analyzed the situation like this when reporting to CEO Nadella:

It was Jakub Pachocki, not Ilya, who really drove the research breakthrough. His contribution was so significant that Ultraman made an exception and promoted him to lead the core model development direction.

Since assuming this important role, Jakub's research progress has accelerated significantly - especially in recent weeks, Jakub has made shocking breakthroughs.

Ilya must have been extremely tormented by this: his former subordinate not only became a leader, but more importantly, this person actually overcame the core problem that had troubled him for many years but still had no solution.

For a proud genius like Ilya, it is not difficult to imagine the impact on his psychological defense. At this moment, even the smartest minds in Silicon Valley cannot escape the frailties of human nature.

So, on November 17, 2023, the OpenAI board of directors announced a blockbuster that shocked the technology world:

Announced the dismissal of co-founder and CEO Altman and kicked him off the board of directors.

In its official statement, the company stated sternly:

Altman's failure to be consistently candid in his communications with the Board hindered the Board's ability to carry out its responsibilities. The board no longer has confidence in him continuing to lead OpenAI.

According to media reports such as the Wall Street Journal, Ilya Sutskever, the then chief scientist of OpenAI, played a key role in the dismissal.

He used his influence with the board of directors to get them to accept his story, and Altman was fired.

Just hours after Altman's decision to be dismissed was announced, OpenAI co-founder Greg Brockman resigned as chairman in anger. Subsequently, technical backbones of the "Polish Legion" such as research director Jakub Pachocki, core researchers Aleksander Mądry and Szymon Sidor also resigned in protest.

In the end, this "coup" did not last long and ended with Ultraman officially announcing his reinstatement and regaining the CEO reins.

At a subsequent all-hands meeting, Sutskever tried to defend the decision, insisting it was not a "hostile takeover" but an effort to protect the company's mission.

But after this "removal storm", the rift between Ilya and Ultraman has become irreparable.

In an email to Nadella, Kevin Scott attributed Ilya’s dissatisfaction with Ultraman to two issues, which have been brewing within OpenAI for a long time.

The first issue is the competition between the research department and the application department for computing resources (GPU); the second issue is the promotion of Jakub's position.

Kevin Scott’s email to Nadella

Computing power “resource civil war”

With the popularity of ChatGPT, the "resource civil war" between OpenAI application departments and research departments competing for GPUs is inevitable.

The success of the application department (ChatGPT team) directly led to the tilting of the company's internal resources.

The precious GPUs that were originally used by the research department to explore the future were largely diverted to serve those C-end users who were eager to chat. The computing power needs of the research department, which is responsible for training new models, could never be met.

Ilya needs massive computing power to verify his idea of "superintelligence", which is limited.

Ultimately, Altman promoted Jakub to lead OpenAI’s core model development.

Kevin Scott commented in an email: "From a CEO's perspective, Ultraman's decision to promote Jakub is completely correct."

OpenAI has started commercialization attempts in 2019 and has gradually launched commercial products. The allocation of computing resources must also serve this strategic direction. If talent cannot be converted into products, it will be ruthlessly abandoned.

Compared with the master Ilya who is engaged in pure research, the practical apprentice Jakub seems to be more adaptable to this new game rule.

After leading the move to remove Ultraman OpenAI as CEO in 2023, Ilya announced his departure from OpenAI in March 2024. He also posted a photo with Jakub and others on the X platform.

Is OpenAI better off after losing Ilya?

Jakub left a message under Ilya’s tweet

Jakub left a message under Ilya's tweet, calling him his mentor and excellent partner, and said he would miss working with Ilya.

Ilya's sad departure is not only a battle for position, but also the collapse of Ilya's "prophet image".

Some netizens bluntly called Ilya "arrogant" and unearthed past media reports about Ilya's "mystery".

According to the Atlantic Monthly, employees chanted "Feel AGI! Feel AGI!" This slogan was spearheaded by Ilya himself.

He even commissioned a wooden statue to represent “misaligned” AI that was contrary to human interests, and later burned it.

In short, instead of focusing on actually advancing AI technology in a scientifically rigorous way, some board members seem to be making strange metaphysical claims.

Sam Altman

However, after forcing Ilya away, will OpenAI get better?

social media comments

Netizens criticized Ultraman for being "inconsistent with human interests and only pursuing profit maximization" and inserting advertisements into AI output. They even said that Ultraman's brief dismissal in 2023 should have been the "best ending."

Some netizens alluded to OpenAI’s “stagnant progress” in model development. Since June 2024, OpenAI has encountered obstacles in large-scale model pre-training, and there has been no major breakthrough in more than a year.

social media comments

Some netizens pointed out that OpenAI has not launched a new cutting-edge model for more than a year. ChatGPT users are turning to Gemini. In the long run, OpenAI has lost its competitiveness.

With Jakub officially taking over as the chief scientist of OpenAI in 2024, Ilya sadly left the scene, and the "Polish Legion" completely came to the forefront.

When the smoke of infighting clears, the market will never pay for Machiavellian stories, but will only vote for disruptive products.

Facing the strong performance of competitors such as Google and Anthropic, whether the "Polish Legion" can regain its former leading position after OpenAI sounded the "red alert", they must come up with a shock like GPT-5 as soon as possible to break the doubt.

References:

https://x.com/trajektoriePL/status/2018242446577086589

https://officechai.com/ai/ilya-sutskever-was-unhappy-that-jakub-pachocki-was-making-more-research-breakthroughs-and-had-been-promoted-microsoft-cto-kevin-scott-during-openai-coup

https://www.geekwire.com/2026/the-microsoft-openai-files-internal-documents-reveal-the-realities-of-ais-defining-alliance