OpenAI's smart revenue push, self-driving wars, and what to make of our new AI king

Welcome to Cautious Optimism, a newsletter on tech, business, and power.

• 📈 Trending Up: Robotics … spine … one-man government … the TikTok ban … bird flu …

  • Good wishes for Sid over at GitLab who is fighting cancer and is stepping back from the CEO role to get well. (I’ve interviewed Sid a time or two, including one memorable occasion when he asked if we could livestream a background call to YouTube. Good times.)

📉 Trending Down: Bitcoin, back under the $100,000 mark … clean government in Boston … memecoins … the Euro …

• Democracy watch: If any deci-billionaire and greater can drop a quarter billy into an election and sway it — how long until our electoral system is just dueling billionaires?

OpenAI wants your money

News that Tinder had introduced a $499 monthly tier raised eyebrows last year. Chatter aside, the move made made sense: If you can charge your power users more, do so. If not you’re just leaving money on the table.

Enter OpenAI, which dropped a new, $200 per-month subscription option this week. People were less than pleased at what was perceived to be much-anticipated features being held back for the new expensive offering, leading to the company’s CEO Sam Altman working to quiet the waves:

Folks had read OpenAI’s news as saying that folks who paid a pedestrian $20 per month — myself included, because my dad depends on my OpenAI account — would not get access to the newly unleashed o1 model, previously in preview. Not so, Sam says, you too will get it.

Kinda. OpenAI is clear that for $20 you get “limited access to o1 and o1-mini,” and folks who pay 10x as much get “access to o1 pro mode, which uses more compute for the best answers to the hardest questions.” So, you mostly get some o1 for $20, and lots more and better for $200. Fair enough.

I might be slightly sulky that I would have to start paying a cellphone bill for OpenAI instead of a Netflix debit to get Big Boy o1, but from the perspective of OpenAI not wanting to lose money forever, it’s a brilliant move.

There’s been ample analysis indicating that while Anthropic makes most of its money through its API — Anthropic’s Clause 3.5 Sonnet model is popular for a reason — OpenAI is, in contrast, largely a subscription game. The $20 that I, and so many others pay, is real dollars. Between 73% and 85% precent of OpenAI revenue, depending on who you read.

OpenAI was previously reported to be on target for $3.7 billion worth of revenue this year while losing around $5 billion. For fun, let’s use 80% as our indicator of the portion of OpenAI that comes from ChatGPT subs. That’s $2.96 billion, which we’ll round to $3 billion.

If 15% of those subscribers swapped from $20 per month to $200, how much more revenue could that unlock for OpenAI?

For fun, $3 billion divided by 240 [12 months * $20] works out to 12.5 million people. 15% of that number is 1.875 million. Those folks monetizing at $2400 per year instead of $240 works out to $4.5 billion. Now we’d need to subtract their prior $450 million worth of revenue before adding that figure to OpenAI’s 2024 revenue tab, but you can see how much a higher priced tier can boost the company’s growth. [Note: The math in this paragraph is at best directionally accurate. Do not trust it as anything more than a little thought experiment.]

CNBC reported that OpenAI “expects to bring in $11.6 billion in sales” in 2025. One great way for it to get there is to charge more. And now it is.

The self-driving wars are hotting up

The newsletter is late today because we had Deedy from Menlo and Coffeezilla on the podcast, which required more prep than usual. One story I did prepare for that we didn’t get to due to the aforementioned interviews was news that Waymo is coming to Miami with Moove (more on Moove here from TechCrunch).

Here’s the key quote (emphasis added):

We'll first collaborate with our fleet partner Moove in Phoenix, where they'll begin taking on the management of our fleet operations, facilities, and charging infrastructure. This strategic partnership is designed to maintain and enhance the clean, consistent experience that our riders have come to expect and rely on. In both Phoenix and eventually Miami, Waymo will continue to offer our service through the Waymo One app, and remain responsible for validation and operation of the Waymo Driver.

Waymo handing off the ticky-tacky elements of running a self-driving program makes sense. It’s not what Waymo is good at; and it’s low-margin work at best. So, Waymo should focus on the tech, and let the experts handle fleet management.

The move does indicate that Uber-Waymo is not a partnership that will trump all others; in turn, that fact implies that Uber would do well to keep spreading its self-driving bets. And it is, announcing this week that it’s teaming up with recent IPO WeRide in Abu Dhabi:

Starting today, Uber riders in Abu Dhabi requesting UberX or Uber Comfort may be matched with a WeRide AV for qualifying trips. […] The launch represents the first time autonomous vehicles (AVs) are available on the Uber platform outside of the United States, as well as the largest commercial robotaxi service outside the U.S. and China.”

The Middle East is currently trying to turn petrodollars into technology winnings. That’s why Abu Dhabi is in play here. Not that I think pushing into tech is a bad idea by wealthy petrostates. It probably is. But it’s worth keeping tabs on how much money in tech today is flowing thanks to sovereign wealth funds controlled by monarchs — and I do not approve of monarchy. (Recall G42 and Cerebras?)

Sadly none of the above news involved a major company bringing self-driving to Providence, so I must continue to wait with a sad face.

And then there’s the Sacks thing

I was not shocked with David Sacks got a gig with the White House for 2025. The incoming “White House A.I. and Crypto Czar” will have a lot on his plate, and that’s not even getting into overlapping realms of responsibility with other Federal organs. The SEC is going to have its own crypto posture, of course.

There’s potential problems aplenty with putting Sacks in the czar chair. His firm has lots of investments in AI (xAI, Ragie.ai, Glue, Glean, Copy.AI) and crypto (Dune, BitGo, Bitwise). Not self-preferencing for a businessperson is going to be weird. And high-profile to boot.

But divestment issues aside, it’s worth keeping in mind that Silicon Valley writ large is run by powerful humans. By that I mean people with a lot of money who compete, and, at times, hate one other with a passion. Recall earlier this year that Sacks et al got into a scrap with Paul Graham and Matthew Prince. Potential issues of bias won’t merely hinge on portcos, but also personal axes that might be in wanting of a little grinding.

And when Sam Altman congratulated Sacks, Musk laughed at him:

That’s because, and bear with me here, OpenAI has advocated for government AI regulations. Musk, a Trump acolyte, is also heading into a government role of some sort, and has insane pull with the next POTUS — at least for now. But if you are in favor of AI regulation and want it to get full shrift, seeing Sacks’ fellow PayPal mafia member straight up mock you to your face on the social media megaphone that he bought and is using to train his own AI foundation model company has to be, well, pretty shit.

That Musk was in favor of California’s much-hated AI regulation bill is merely ironic fodder for the dead horse I am beating.

More Monday. — Alex