Last week there was only one piece of AI news that truly mattered. Gemini 3 challenged the assumption that’s dominated the industry for the last two years: that OpenAI is unbeatable. Shortly after, an alleged internal memo from Sam Altman surfaced online, in which the head of OpenAI admits the company is in for a period of “rough vibes” and short-term pressure. We don’t know if this note is genuine, but it would be better for the creators of ChatGPT if it were. It would mean that OpenAI is taking the issue seriously and has a plan to deal with it.<\/p>
For a long time, the balance of power seemed stable. For the general public, AI = ChatGPT. OpenAI was setting the pace, while Google and Meta were trying to keep up, and other companies were carving out their niches, hoping to gain an edge in specialized applications. Each new version of GPT reinforced this narrative. It was OpenAI dictating the AI landscape, with the rest of the market following its developmental path.<\/p>
However, with the launch of Gemini 3, there was almost a “tectonic” shift. Benchmarks that had been even or slightly leaning towards OpenAI for a long time suddenly started showing a clear advantage for Google in areas that had been the GPT’s stronghold: complex reasoning, multimodality, visual data interpretation, and quasi-cognitive tasks that are now so intensely tested in the new AGI benchmarks.<\/p>
The scale of differences is so significant that it can’t be ignored or swept under the rug with PR narratives about methodological differences. Gemini 3 has taken the lead unquestionably, and in crucial areas for the future of AI. For the first time in a while, the market felt that the leader’s board could change, and it all happened almost overnight.<\/p>
Shortly after, information about Altman’s memo to employees surfaced online. The document isn’t catastrophic \u2014 it lacks a defensive tone but includes phrases that clearly signal a phase change. Altman talks about “rough vibes,” “temporary headwinds” and the need to focus on “very ambitious long term bets.” It sounds like a declaration of stepping back from the short-term race while also a message about building something significantly bigger. It’s not a letter from someone afraid of losing their position. It\u2019s a letter from a leader telling his people: now we’re facing our toughest challenge, and part of that is accepting the fact that we have temporary problems.<\/p>
Increasingly more data suggest that OpenAI’s real problem lies at the very foundation of their model creation process. The industry refers to a “pre-training wall” that OpenAI has apparently hit, as they try to squeeze more out of existing methods than they are capable of delivering. In practice, this means each new batch of computing resources yields diminishing returns in terms of quality. The company has a massive infrastructure, but its data pipeline, cleaning methods, optimization schedules and training layers are beginning to slow down further development. It seemed like there was no solution to this problem…<\/p>
Meanwhile, Google showcased something that looks like a return to its roots. A refined pre-training process, massive data sets, optimized infrastructure and tight integration of models with service products demonstrated that the classic development path can still produce huge quality leaps. This hasn’t been an architectural revolution. It’s an organizational revolution, and it’s this difference that really stung OpenAI.<\/p>
In the context of Altman’s memo leak, the name Shallotpeat also comes up. It’s a project that reportedly aims to fix issues that have been building up in the foundations of OpenAI’s training processes for a long time. The name, light and almost humorous, perfectly captures the metaphor of its message: change the groundwork before putting up another skyscraper. Shallotpeat is said to improve data purity and structure, enhance pipelines, resolve quality issues and prepare the field for a significant leap in capabilities.<\/p>
If this succeeds, it won’t just be an answer to Gemini 3. It will be a whole new factory, not just a new model.<\/p>
It’s worth noting that Altman has long been talking about goals that go beyond the classic benchmark competition. In many of his appearances, he emphasizes that the real breakthrough won’t be GPT 6 or GPT 7, but the automation of AI research. Models that write code for other models. Models that design training experiments. Models that can analyze their own mistakes and generate sets of corrections. The winner won’t be the company with the best model, but the one with the best process for creating them. That’s exactly why OpenAI has started investing in organizational change, not just parametric adjustments. Shallotpeat seems to be the first step in this evolution.<\/p>
So is OpenAI really losing its lead? In one dimension, yes. Benchmarks are now in Google’s favor. However, the rest of the picture is more complex. OpenAI still boasts the largest user platform, the most extensive tool ecosystem, the strongest presence in business integrations and the most active developer community. They might lose the sprint, but they are winning the marathon.<\/p>
For the market, this marks the period of greatest uncertainty since the launch of GPT 4. If Google keeps up the pace, AI-based tools will increasingly migrate to Gemini, and companies will start planning their strategies around DeepMind products. If OpenAI comes back with a breakthrough that could spring from an improved pipeline, we might witness another leap similar to the shift from GPT 3.5 to GPT 4. However, the most likely scenario is one where both companies speed up side by side, leading to the most intense technological rivalry of the decade.<\/p>","protected":false},"excerpt":{"rendered":"
Gemini 3 instantly showed that market dominance is never guaranteed forever. Does OpenAI have a chance of turning a failure into a success?<\/p>\n","protected":false},"author":465,"featured_media":16324,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_lock_modified_date":false,"footnotes":""},"categories":[813],"tags":[],"popular":[],"difficulty-level":[38],"ppma_author":[892],"class_list":["post-16333","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-news-2","difficulty-level-medium"],"acf":[],"authors":[{"term_id":892,"user_id":465,"is_guest":0,"slug":"kmironczuk","display_name":"Krzysztof Miro\u0144czuk","avatar_url":{"url":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/10\/awatar-2.png","url2x":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/10\/awatar-2.png"},"first_name":"Krzysztof","last_name":"Miro\u0144czuk","user_url":"","job_title":"","description":"Od lat zajmuj\u0119 si\u0119 nowymi technologiami w biznesie, edukacji i codziennym \u017cyciu. W centrum mojej uwagi pozostaje cz\u0142owiek \u2013 i to, by technologia wyr\u00f3wnywa\u0142a szanse, zamiast tworzy\u0107 bariery."}],"_links":{"self":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/16333","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/users\/465"}],"replies":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/comments?post=16333"}],"version-history":[{"count":1,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/16333\/revisions"}],"predecessor-version":[{"id":16334,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/16333\/revisions\/16334"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media\/16324"}],"wp:attachment":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media?parent=16333"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/categories?post=16333"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/tags?post=16333"},{"taxonomy":"popular","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/popular?post=16333"},{"taxonomy":"difficulty-level","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/difficulty-level?post=16333"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/ppma_author?post=16333"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}