May 18, 2024

Good morning!

  • ☎️ Reminder: Feel free to send me tips or comments by replying to this email or on X: @imkialikethecar. (Or ask me for my Telegram or Signal number.)

Today's Smart Brevity™ count is 789 words — a 3-minute read.

1 big thing: Into the Elon-verse

Illustration: Shoshana Gordon/Axios

Investing in Elon Musk's xAI is an opportunity to get involved in "the Muskonomy" of ventures fronted by one of the world's wealthiest men.

Why it matters: xAI is said to be raising as much as $6 billion to catch up to rivals like OpenAI and Anthropic, and it needs a compelling story for investors.

The big picture: xAI's close relationship with X — formerly known as Twitter — is key to its potential success and one aspect of the "Muskonomy," according to the deck prepared by Brookfield late last year and seen by Axios.

  • "The fuel for AI is data," notes one slide. "X represents the world's most timely dataset of real-time human conversations, interests, and trends for xAI to form its intelligence. X's dataset will also become valuable as it progresses on its 'everything app' vision."
  • In addition to the sheer size of data the AI company can access from X, the deck points out that as users continue to generate new posts, xAI will have a constant feed of new data to improve its model.
  • The deck also highlights how xAI will be able to tap into video data from automaker Tesla and have "strong alignment to human safety" by partnering with Neuralink, Musk's brain chip startup.
  • Brookfield declined to comment to Axios, and X did not immediately respond to a request.

Between the lines: It's not yet clear what other social media data xAI plans to use to feed its models, but having baked-in access to one major network is significant.

  • Most of its competitors have resigned themselves to signing licensing deals with social media companies to use their data — just this week, OpenAI signed one with Reddit, for example.

Thought bubble: The deck is a long and detailed way of making an argument Musk's supporters have long bought into: The billionaire will turn anything into a success, so best to buy a seat early before the rocket ship takes off.

The intrigue: A number of blue chip investors will surely back xAI, but a number of smaller backers also seem to be piling into the company, likely via special purpose vehicles, according to various SEC documents.

  • Axios reached out to a few investors named in those filings, but none replied to confirm involvement.

The bottom line: In the AI wars, everyone is looking for a moat — and xAI's might be Musk.

2. What they're saying: Perplexity AI CEO on fundraising

Illustration: Aïda Amer/Axios

At Axios' BFD event Tuesday, Perplexity CEO Aravind Srinivas dismissed recent rumors that the company is raising a large new funding round.

Yes, but: He had this to say about AI companies' seemingly never-ending fundraising:

There are very few people who really understand even prompt engineering, making these models do what you want them to do, making it do it reliably at scale.… This requires a lot of tacit knowledge that doesn't exist in the wider engineering community right now. It's not yet commoditized.

So, these people are expensive and all their salaries got inflated because OpenAI started paying them a lot, and paid them through equity when they used to be valued at like $15 [billion] to $20 billion.

But now that they're valued more close to $100 billion, their salaries are all like 5x, 6x what they used to be. Now that becomes their market value, and now when other companies try to compete with OpenAI for talent, they're also willing to pay that much.

So is Perplexity playing this game?

I try not to. At least go for talent that wants to work at earlier-stage companies and not like just optimizing for money, but in general, at some point you do need to hire very good, talented people.

So that's why you do need to raise more money to pay people really well, or to have a high-enough valuation that, like, you don't need to dilute their entire option pool just to hire one or two people.…

And this the other argument of building a big cluster — and so [graphic processing units] are expensive.… Out of the $6 billion Musk was raising and he's already committing to spend $10 billion on Oracle — he's committing more than the amount he's even raising. That's a trend.… There are many startups where they raised $70 million but they committed like $300 million.

📚 Due Diligence

  • OpenAI dissolves high-profile safety team after chief scientist Sutskever's exit (Bloomberg)
  • Perplexity CEO Aravind Srinivas takes shots at Google (Axios)
  • AI eats the web (Axios)

🧩 Trivia

Perplexity is not the first (and hopefully not last) buzzy AI startup we've had on the Axios BFD stage.

  • Question: Which company did we interview that touted ambitions to be a large public company, only to announce months later its acquisition by a tech giant? (Answer at the bottom.)

🧮 Final Numbers

The chart shows the annual deal activity in AI and machine learning from 2015 to 2023. It indicates a steady increase in both deal value and deal count over the years, with a peak in 2021 at $142.9 billion and 9,645 deals. Total deal value for 2023 was $90.9 billion with 7,238 deal counts.
Data: PitchBook; Note: 2023 data is as of Dec. 31; Chart: Axios Visuals

🙏 Thanks for reading! And to Javier David and Brad Bonhall for editing. See you Monday for Pro Rata's weekday programming, and please ask your friends, colleagues and AI investors to sign up.

Trivia: Inflection AI