Axios AI+

July 22, 2025
And, I'm back. Thanks so much to Megan, Scott and the gang for keeping y'all in the know while I was off. Today's AI+ is 1,214 words, a 4.5-minute read.
1 big thing: AI's anything-goes era
The AI industry is getting nothing but green lights in all directions — now it needs to deliver on its promises.
The big picture: AI makers are getting everything they have ever asked for or could possibly want.
1. No limits: More money, energy and resources are flowing into AI than any other industry has received in such a concentrated timespan.
- Four companies — Alphabet/Google, Microsoft, Meta and Amazon — expect to spend over $300 billion this year on AI, while private investors and governments pour hundreds of billions more into AI infrastructure.
- Public and private projects are rushing to supply the vast energy inputs AI development and use requires.
- A Pittsburgh summit featuring President Trump last week made clear that the emphasis will be on fossil fuels and nuclear, with little regard for climate concerns or environmental costs.
2. No rules: In the second Trump era, the U.S. has dropped any pretense of building regulatory guardrails around AI.
- Trump's AI "action plan" coming this week will instead promote speedy deployment to counter China.
- The "doomer" camp's fear that runaway superintelligence might destroy humanity is no longer even a part of the policy conversation.
3. No arguments: Gung-ho CEOs and businesses are pushing AI use on sometimes resistant workforces and a skeptical public, telling hesitaters to get on the AI train or get left behind.
- The phenomenal popularity of ChatGPT and its competitors suggests there's tons of demand for these tools.
- But surveys also show the U.S. public fears and distrusts AI and favors a more careful approach to its adoption.
4. No doubts: Business leaders and policymakers have successfully sidelined critical questions about harms from AI bias and misuse, privacy violations and appropriation of intellectual property.
- If you point out that today's AI is inefficient, untrustworthy and unstable, they will say "just wait till next year's model."
- The new technology is being promoted as a race to "superintelligence" where the winner will reap enormous — but unspecified — rewards.
Why it matters: All these green lights are flashing at a critical juncture in the development of AI.
- Today's large language models are changing at high speed as giant data centers come online, researchers apply new techniques, companies plug the new tools into their workflows and software developers build bridges between AI and the existing digital world.
- That means AI's formative era is right now — and the technology is developing with almost total freedom.
Our thought bubble: AI makers liken the process of training models to raising children. If that's true, the technology is growing up as a fabulously rich kid in a wildly permissive household.
Yes, but: Now that the world has said yes to every ask the AI makers have made, it's "put up or shut up" time for the new technology.
- AI champions predict a utopian cornucopia of benefits just beyond the horizon: Massive boosts to productivity and economic growth! Miracle drugs and cures for cancer! Personalized tutors for every student and Ph.D.-level interns for every company!
"If we get this right, AI can give everyone more power than ever," Fidji Simo, OpenAI's new CEO of applications, wrote in a blog post yesterday — though she also admits that these benefits "won't magically appear on their own."
- In fact, though, today they remain largely hypothetical, while the technology's costs pile higher.
The bottom line: ChatGPT will hit its three-year anniversary this fall.
- Since its arrival, AI leaders have been confidently forecasting that the massive breakthrough of a "singularity"-style superintelligence is a mere year or two away.
- This technological event horizon has remained steadily distant as years have passed. But the industry can't play the "nirvana is just a few years off" card forever.
2. Google and OpenAI fight for math gold
Google DeepMind and OpenAI both achieved gold medal-level performance at this year's International Mathematical Olympiad (IMO) over the weekend — but only Google officially entered the competition.
The intrigue: Google DeepMind's model results were certified by the IMO, but OpenAI released its own results first, highlighting the speed and urgency in the race to build the best model for math and reasoning.
- The IMO is an elite math competition for high school students, drawing participants from over 100 countries. It was held in Australia this year.
The big picture: OpenAI didn't enter the competition, but evaluated its model on the 2025 IMO problems, after seeing the model's performance on related tasks.
- Researcher Alexander Wei shared OpenAI's results on X on July 19.
- The model abided by the same rules as human contestants, including two 4.5-hour exam sessions using no internet or other tools.
- Google announced Monday that an advanced version of Gemini Deep Think solved five out of the six IMO problems perfectly, earning 35 total points.
- That's the same score OpenAI announced.
Between the lines: The results from both companies show how far general-purpose models have progressed in solving advanced math problems and delivering the answers in natural language proofs.
- Earlier AI wins in Go and Poker came from models trained just for those games.
- The new high-performing models are general purpose models, the same ones they train for language, coding and science.
Why it matters: AI models are unusually difficult to benchmark because of the speed at which the tech is moving and the lack of a standard benchmarking system.
Zoom out: Both models are experimental and won't be released to the public "for a while," OpenAI says. "Many months," according to a post on X from OpenAI CEO Sam Altman.
- The model that competed in the IMO is "actually very close to the main Gemini model that we have been offering to people," Google DeepMind senior staff research scientist Thang Luong told Axios.
- "We hope that this will empower mathematicians so they can crack harder and harder problems."
What they're saying: "When we first started OpenAI, this was a dream but not one that felt very realistic to us," Altman said on X.
- "Our leap from silver to gold medal-standard in just one year shows a remarkable pace of progress in AI," Google wrote on its blog.
Yes, but: Both Google and OpenAI praised the high school students participating in the Olympiad and were careful not to frame the competition as a bots vs. humans cage match.
- The purpose of the IMO is to promote the "beauty of mathematics" to high school students and to encourage them to go into the field, Junehyuk Jung, associate professor at Brown University and visiting researcher at Google DeepMind, told Axios.
- Jung was a participant in the IMO 22 years ago.
Google waited for the IMO to officially certify the competition results rather than release its results over the weekend out of respect for the students in the competition, Luong said.
- In his post on X, Wei pointed out that OpenAI employs many former IMO participants and called them "some of the brightest young minds of the future."
3. Training data
- OpenAI and SoftBank are struggling to get their Stargate project up and running amid challenges and differing views. (WSJ)
- Anthropic plans to seek investments from UAE and Qatar, overcoming qualms about enriching "dictators," per a confidential memo. (Wired)
- PayPal and Venmo added AI alerts to warn users about scammy "friends and family" payments.
4. + This
It's been around for a decade, but I just learned there's a high school in Pennsylvania that bears a striking resemblance to the Millennium Falcon.
Thanks to Scott Rosenberg and Megan Morrone for editing this newsletter and Matt Piper for copy editing.
Sign up for Axios AI+





