OpenAI’s ‘Code Red’ Shows the Power of Perceptions in the AI Race

The Week in Short

‘Code Red’ at OpenAI as Google & Anthropic surge. David Sacks v. NYT misses the big picture. Unicorn herd gets richer. Kalshi & Polymarket go gambling. US government invests in laser startup. Harvey, Curative & Eon raise fresh funds. Apple shakes up AI leadership. Meta to cut Metaverse. FDA chaos imperils biotech. Trust & safety workers in Trump crosshairs.

GiveDirectly

Newcomer Contribution Matches Are Still Live

Many thanks to our subscribers for supporting our @GiveDirectly campaign. Substackers have raised about $1 million to date for cash assistance to alleviate extreme poverty. You can catch up on it here and we’re still matching your donations at givedirectly.org/newcomer.

The Main Item

As the AI Narrative Shifts to Google & Anthropic, Sam Altman’s Media Mastery is Put to the Test

With AI investment holding up the global economy and the foundation model race continuing at a fevered pace, CEOs are leaning into the adage from the prescient Google researchers whose 2017 paper kicked off the current boom: attention is all you need.

Over the past couple of weeks, the attention has been going to Sundar Pichai and Google, whose new Gemini 3 model won wide praise and topped the leaderboards that measure performance. And it’s been going to Dario Amodei and Anthropic, whose “we’re safer” message — along with a top-notch model that’s especially good at coding — is resonating with business customers. Anthropic is now starting down the road to an IPO, which promises lots of publicity too.

Where does that leave Sam Altman and OpenAI, which have dominated the AI storyline to date? He declared a company-wide “Code Red,” which involves delaying some initiatives and shifting resources to an all-out effort to further improve its core models. Altman also praised Gemini and told employees, “I expect the vibes out there to be rough for a bit.”

That might sound like a CEO giving a sober assessment of the state of play and rallying the troops to do more, faster. But in an AI race with unprecedented multi-trillion-dollar stakes, nothing is quite so simple.

In truth, as Newcomer’s Tom Dotan wrote back in April, Google, with all of its formidable assets, was never very far behind. Nor is it currently very far ahead. Anthropic too has always been essentially neck-and-neck with OpenAI on the core technology.

The capabilities of the