What we can infer from OpenAI’s latest hire
Also: We're in the age of the age-gated internet
Welcome to Cautious Optimism, a newsletter on tech, business and power.
Wednesday. The Fed cut! The expected 25 bips, but it’s still nice to wrap the year with a little juice for the system. If you are hoping for even more rate cuts in the months to come, perhaps temper those expectations a little. To work! — Alex
📈 Trending Up: Housing for the homeless? … no shit … internal computing wars … self-driving … open-source AI … Amazon in India … open-weight AI … executive accountability … geothermal power … Naveen Rao …
📉 Trending Down: Calendrical competence … affordable groceries … dependence on China’s rare earth industry … AI prices in India …
Things That Matter
Fly Me to the Moon: Reporting that space launch and satellite Internet company SpaceX intended to go public in 2026 is getting more specific. After the WSJ reported that the company was “weighing a potential initial public offering in 2026,” Bloomberg dug in and managed to get more details: SpaceX could raise as much as $30 billion in an IPO “as soon as mid-to-late 2026” at a $1.5 trillion price tag.
Backing that sky-high valuation are revenues of $15 billion this year, Bloomberg writes, that are set to increase to “between $22 billion and $24 billion in 2026, [with] the majority of sales coming from Starlink.”
If SpaceX beats its expectations and reaches revenues of $25 billion next year, a $1.5 trillion price tag would value it at around 60x revenues. That’s very expensive. SpaceX is such a revered name in tech circles that it may be able to name its price, but it does face increasingly sophisticated competition, and rivals around the world are racing to replicate its Starlink service.
Regardless, if we do get a SpaceX IPO next year, I’ll consider 2026 a solid news period even if nothing else happens.
Palantir wants to help the US build ships: With a new $448 million contract, Palantir is working on ‘ShipOS’ with the Navy, software that will offer “predictive work on supply chain management, where every single one of the bits involved is a special-purpose bit,” CEO Alex Karp told Axios.
The goal? Get American military shipbuilding — above and below the waves — out of the muck, where it is, let’s be frank, currently mired. Examples abound if you care about such things.
Will Palantir’s ShipOS save the Navy? Your guess is as good as mine here, but at least we’re trying.
The age-gated Internet: Starting today, Australian kids under the age of 16 are banned from a host of social media platforms, including Instagram, X, YouTube, Reddit and Twitch. The nation doesn’t expect perfect performance on day one, and instead expects to ferret out kids who get past age-verification technology, like facial scanning.
Arguments for limiting youth access to certain technologies are picking up momentum around the world, and they range from the obvious (limiting access to smartphones during school hours) to draconian (see China, Iran or North Korea).
Australia’s social media ban is more extreme than other national experiments in age-gating the Internet, but it isn’t unique. Some American states are also poking at limiting kids’ access to social media, though industry groups are opposing the moves in court.
Will efforts to block kids from accessing social media work? Not entirely. But data from the UK indicates that making changes to how the Internet works (age verification for accessing adult content) does impact citizen behavior (lower adult content is used in the nation). Similarly, making it harder for all kids under a certain age to make a social media account will limit activity, even if it won’t block all kids all the time.
The first age of the Internet is long behind us. Information is paywalled, and blocs of the Internet are being closed off from a material percentage of the population. The trend of closing off access and information only appears to be accelerating — the boom of AI has led to many websites and companies working harder than ever to defend their own material.
There are major privacy concerns regarding all of the above, and losing our ability to be anonymous online would be quite destructive, especially in nations where free speech is a distant ideal. But it appears that this is where we’re heading, because the people seem to want it. Polling from 2024 indicated that, at the time, “77% of Australians backed the under-16 social media ban, a significant increase from the 61% support found in an August poll prior to the government’s official announcement.”
What we can infer from OpenAI’s latest hire
During my time in Startup Land, certain companies’ former employees had dubious reputations. If, say, you hired a bunch of folks from Uber, you’d import the company’s work culture, which at the time could lead to cultural clashes as Uber employees were known for sharp elbows and dog-eat-dog vibes. You might lose some high performers accustomed to collaboration, not sparring, for example. Former Amazon staff had a similar reputation.
While it’s unfair to pick on any single company, the lesson was that hiring workers from other companies would bring part of that company’s culture, too.
This is why, when OpenAI hired now-former Instacart CEO Fidji Simo, I expected the company to pursue advertising and integrate commercial partners into ChatGPT. This is less a point about culture and more to do with product, but culture is a product, and products are the result of culture.
Similarly, now that OpenAI has hired Slack CEO and longtime Salesforce exec Denise Dresser as its Chief Revenue Officer, we can infer what we might see from OpenAI in the future by contemplating her background.
I’ve worked with a bunch of ex-Salesforce folks; ask me about it over coffee sometime.
Salesforce is great at building software that businesses pay for, and has run a strong M&A playbook en route to becoming one of the largest tech companies in the world — Slack, a Salesforce unit, was once the fastest-growing software product in the world, for example.
Dresser’s experience is therefore a good fit for OpenAI, which is making noise about its enterprise penetration. Just expect what she cooks up to rhyme with what Salesforce learned from its ownership of Slack: namely, protected data, and more expensive, finely-grained enterprise packages.
This is good, given that OpenAI needs the revenue. But I do wonder how a company that started life as a lab would survive the transition to a more traditional corporate leadership without avoiding a lot of internal discontent.
Then again, perhaps that’s precisely what OpenAI needs to regain its edge in the AI race.