Good morning. I had the privilege of moderating a space dinner last night in metropolitan Los Angeles.
Alas, not a dinner in space—someday!—but a meal and discussion featuring an all-star group of aerospace and defense executives. Turns out it’s a big and fast-growing business in La La Land, fueled by commercial ambition, national security, and (naturally) copious venture capital.
For your benefit, a haiku of the proceedings:
Strong winds in our sails
This course is not yet sustained
Investors, hold fast
Many thanks to Deloitte for making it happen. Today’s tech news, below. —Andrew Nusca
Want to send thoughts or suggestions to Fortune Tech? Drop a line here.
Breaking convention, SpaceX sets IPO price before roadshow
The magic number for the SpaceX IPO? $135 per share.
With almost 556 million shares on offer, Elon Musk’s rocketmaker (and the world’s most valuable private company) aims to raise $75 billion in its imminent public debut.
But traditionalists would say the timing is all wrong.
Typically a company going public engages in a roadshow where executives pitch their stock offering to prospective investors. Once interest is assessed, and just before the offering itself, a price is determined.
Not so for SpaceX—but then again, Musk has never been one to follow standard protocol.
SpaceX’s CEO, who retains a 42% stake in the company but controls more than 85% of its voting power, stands to become—with a roughly 2% increase in share price—the world’s first trillionaire based on net worth.
The all-important date for SpaceX’s debut is June 12. Its Nasdaq ticker will be SPCX.
And what will SpaceX use all that money for, you ask? AI computing infrastructure via its xAI (now “SpaceXAI”) unit, the expansion of its Starlink satellite network, and R&D for its Starship rocket and planned Mars missions.
Lest you think the Space Exploration Technologies Corp. makes most of its money from space exploration, a useful reminder: Three of every five dollars SpaceX makes comes from Starlink. The rest? AI and space, a buck each. —AN
OpenAI calls for mandatory cyber risk evaluations of AI models
In the wake of a new Trump executive order calling for private AI companies to voluntarily submit their models for cyber risk evaluation, OpenAI says it’d rather make them mandatory.
What’s more, OpenAI would rather the Commerce Department’s Center for AI Standards and Innovation (CAISI) do the evaluating, instead of the Defense Department’s National Security Agency.
“As AI becomes increasingly important, democratic governments—not private companies acting alone—must ultimately determine the rules, safeguards, and accountability mechanisms,” the company writes. “Our view is that decisions about the pace of AI innovation should not be left to any one lab, company, or special interest group.”
OpenAI’s proposal—“a blueprint for a federal framework”—comes as CEO Sam Altman visits Washington for a series of meetings with White House officials and legislators.
Chris Lehane, a veteran political operative turned Chief Global Affairs Officer at OpenAI, told reporters Wednesday that CAISI has relationships with AI companies that the NSA does not.
He also said the White House plan to benchmark certain frontier models was vague. (The order leaves it to NSA director Gen. Joshua Rudd, a longtime Special Forces officer, to determine which models to evaluate.)
It’s safe to assume OpenAI is trying to influence the frontier AI portion of the order, which takes effect in 60 days. We’ll soon see how things evolve. —AN
U.K. says publishers may opt out of Google's AI Overviews
Digital publishers can choose to not appear in the AI Overviews found in Google search results in the U.K., the country’s antitrust watchdog said Wednesday.
The Competition and Markets Authority, or CMA, imposed on Google a so-called conduct requirement that allows publishers to “prevent their content being used to power AI features in search.”
The new rules also allow publishers to opt-out of allowing their content to be used to “fine-tune” AI models.
“It is crucial that content publishers, including news organisations, have appropriate bargaining power over how their content is used,” CMA CEO Sarah Cardell said in a statement.
(The catch, of course, is that publishers who opt out won’t receive any referral traffic at all from AI results. But that’s been the “clickless” problem all along.)
In a blog post, Google said it has “increased the number of inline links directly within responses” and “brought Preferred Sources into AI Overviews and AI Mode” so people can more clearly see where information comes from.
The company says 2.5 billion people use AI Overviews each month; more than one billion use AI Mode. Google Search meanwhile counts more than four billion. —AN
More tech
—China is flooding online job platforms with fake profiles and offers targeting government and military personnel.
—Meta will spin out Supernatural, the VR fitness game, as an independent company.
—YouTube overtakes Netflix in average daily viewing around the world, according to market researcher Digital i.
—Baidu plans to spin off and list Kunlunxin, its AI chip subsidiary, in Hong Kong and Shanghai this year.
—Europol et al dismantle nine organized crime groups engaged in illegal media streaming.
—Suno raises $400 million. The AI music startup is now worth more than $5 billion.
—Legislators call on the U.K. government to end its deal with Palantir.
