1 month ago
22
ARTICLE AD BOX
In January 2018, Elon Musk told Sam Altman that OpenAI was connected a way of definite nonaccomplishment against Google—then discontinue the institution erstwhile his projected hole was rejected. Seven years later, Altman declared a company-wide "code red" implicit Google Gemini's resurgence and Anthropic's endeavor dominance. Musk's structural diagnosis turned retired to beryllium mostly correct. The occupation helium identified remains unsolved, conscionable considerably bigger.
On the past time of January 2018, Elon Musk sent a illustration to Sam Altman, Greg Brockman, and Ilya Sutskever. No lengthy preamble. Just a blunt takeaway: DeepMind and Google Brain were dominating AI research, and
OpenAI
was falling behind. "OpenAI is connected a way of definite nonaccomplishment comparative to Google," Musk wrote.
That warning—detailed successful Sebastian Mallaby's The Infinity Machine: Demis Hassabis, DeepMind, and the Quest for Superintelligence, excerpted successful Colossus magazine—landed astir 8 years ago.
It reads otherwise today.By December 2025, it was Altman sending the alarming memo. He declared a company-wide "code red" to hole ChatGPT, shaken by a surging Google Gemini and a fast-growing Anthropic eating into OpenAI's endeavor lead.
The Wall Street Journal archetypal reported the memo's contents. The roles had wholly reversed.
Google's Gemini comeback spooked OpenAI's leadership
The contiguous trigger was Google releasing Gemini 3 the erstwhile month—a exemplary that cleared manufacture benchmarks and pushed the app to 650 cardinal monthly progressive users by October 2025. Altman told unit OpenAI needed to amended ChatGPT's personalization, speed, reliability, and the scope of questions it could answer, per the WSJ.
This was a striking relation reversal. It was ChatGPT's motorboat successful November 2022 that had primitively acceptable disconnected a codification reddish astatine Google, with Pichai informing his squad that the chatbot could endanger Search itself. Now the panic was flowing the different direction.Anthropic compounded the pressure. By September 2025, it had crossed 300,000 concern customers—up from less than 1,000 conscionable 2 years prior, per CNBC. Its Claude Code and endeavor tools had go the default for bundle developers crossed Silicon Valley.
By March 2026, OpenAI's ain applications main Fidji Simo was telling unit successful an all-hands that Anthropic's occurrence should beryllium a "wake-up call," urging them not to get distracted by "side quests," the WSJ reported.
The governance combat that softly shaped today's AI race
None of this happened successful a vacuum. Mallaby's publication traces the roots of this rivalry backmost to 2015, erstwhile DeepMind co-founders Demis Hassabis and Mustafa Suleyman launched an ambitious—and yet doomed—attempt to physique a information governance operation for AI, internally code-named "Project Mario."The program was a alleged 3-3-3 board: 3 DeepMind members, 3 from Alphabet, 3 independents. Reid Hoffman committed $1 cardinal to backmost a imaginable walkout from Google. The talks dragged connected for 3 years, with Pichai and Google's ineligible main David Drummond moving what DeepMind's founders came to spot arsenic an endless bully cop/bad bull routine. Nothing ever got signed.Meanwhile, the parallel play astatine OpenAI was conscionable arsenic messy. Musk, convinced that Altman was maneuvering for control, grew progressively erratic.
He projected folding OpenAI into Tesla. When the squad refused, helium quit—and stormed retired of an all-hands gathering aft calling an intern who questioned his judgement a "jackass."His January 2018 memo was fundamentally his parting shot. Brockman pushed backmost the aforesaid day, arguing that league papers were a mediocre measurement of progress. "It doesn't substance who wins if everyone dies," helium wrote. But Musk had already made up his mind.
Musk's prediction, and however it aged
Musk's 2018 informing turned retired to beryllium little a parting changeable and much an aboriginal work connected a structural occupation OpenAI is inactive wrestling with today. Google had the probe talent, the compute, and the gross to prolong a agelong war. OpenAI had momentum and a chatbot. For a while, that was enough.But Gemini 3's accomplishment successful precocious 2025 changed the calculus. The exemplary didn't conscionable bushed ChatGPT connected benchmarks—it deployed crossed Google's full ecosystem astir overnight, reaching billions of existing users successful days.
That's the benignant of organisation vantage Musk was pointing astatine backmost successful 2018, erstwhile helium showed Altman the illustration and said the numbers didn't lie.Altman's code-red memo, reported by the WSJ, was successful galore ways an admittance that Musk had a point. OpenAI inactive leads connected play progressive users—800 cardinal arsenic of precocious 2025—but idiosyncratic counts don't wage the bills erstwhile you're burning done currency astatine OpenAI's rate. The institution needs to turn gross to astir $200 cardinal to crook a nett by 2030, per its ain projections. Google, meanwhile, funds its AI ambitions from a hunt concern that prints money.Musk wanted to lick that occupation by folding OpenAI into Tesla. The squad said no, and helium left. Seven years later, the occupation helium identified remains unsolved—just bigger, and with a batch much riding connected the answer.
