Tech
Italian Tech Week 2024
Sam Altman, cofounder and CEO of OpenAI (Stefano Guidi/Getty Images)
Weird Money

OpenAI is in the business of making OpenAI employees rich

OpenAI's stock compensation expense showed that its employees were paid between $400,000 and $2,000,000 in average stock comp through the first six months of 2024.

Jack Raines

Since OpenAI closed its massive $6.6 billion funding round that valued the company at $157 billion, I’ve been wondering how they managed to convince investors that the valuation makes sense. The answer, it turns out, was another large number: $100 billion.

Cory Weinberg over at The Information published an interesting piece breaking down OpenAI’s investor pitch for its most recent fundraise, and some of the numbers they showed investors were astounding. Notably:

Revenue: OpenAI expects revenue to scale from an expected ~$4 billion in 2024 to $100 billion in 2029, which would be a ~90% revenue CAGR over the next five years. While revenue growth over the last year has been explosive (monthly revenue for August 2024 was $300 million, up 1700% since early 2023), growth will become more difficult with size. For example, it’s easier to go from ~$180,000 in monthly revenue to $300 million (as OpenAI did) than it would be to grow from $300 million to $510 billion.

Compute Costs: Ignoring all other operating costs such as salaries, general and administrative expenses, and sales and marketing, OpenAI’s compute costs to train and run its models are expected to be $5 billion this year, compared to $4 billion in total revenue.

Stock Compensation: OpenAI reported stock compensation of $1.5 billion in the first half of 2024, which is around its revenue for that period.

This last point is especially interesting. Two weeks ago, I discussed the curious case of OpenAI’s wave of resignations, as at least nine high-level executives had left the company over the last year. At the time, I pointed out one factor that could be influencing these resignations was that long-time OpenAI employees had the opportunity to sell equity in a tender offer for massive returns:

All of the above-mentioned employees have been at OpenAI since at least 2022, when OpenAI was valued at ~$20 billion, and most of them started even earlier, when OpenAI’s valuation was much lower. In February 2024, they were able to sell some of their stakes in a tender offer at an $86 billion valuation. If you were a long-tenured employee at OpenAI, and you took some chips off the table in that tender offer, you’re rich. And not only are you rich, you are a hot commodity in a hot labor market in the hottest sector in technology right now. You would have no problem raising capital for a new startup or getting paid top-dollar to join another AI startup or a big tech company.

The real question is, if you’re already rich, anyone would hire or fund you, and the company you’ve worked at for years has changed its entire mission statement… why would you stay?

The stock-based compensation stat all but confirms that, yes, OpenAI’s employees have been getting p-a-i-d. For context, Nvidia, a $3.3 trillion company with ~30,000 employees, paid $2.2 billion in stock compensation through the first half of 2024. OpenAI, which is worth roughly 5% of Nvidia, paid 68% of Nvidia’s stock compensation. And the compensation per employee is jealousy-inducing.

In November 2023, OpenAI had 770 employees. According to employee contact database RocketReach, OpenAI now has 3,726 employees. With $1.5 billion in stock compensation paid out in the first half of 2024, the average employee earned between $400,000 and $2,000,000 in stock-based compensation in that six-month period, depending on headcount over the course of that period. Additionally, OpenAI’s CFO confirmed that, as with the February tender offer, employees would again be able to sell shares after this funding round. So, no, we shouldn’t be surprised that OpenAI employees are resigning. They’re millionaires with willing buyers of their shares.

More Tech

See all Tech
tech

Meta will begin using your AI chats to target you with ads

When Meta rolls out a new product, the company usually waits until that product has a billion users before turning on the ads.

In May, Meta announced that Meta AI has crossed that threshold, saying that more than 1 billion people are using the product every month. Today, Meta announced that it will begin using your conversations and messages with Meta AI to personalize your recommendations and the ads you see.

Meta currently monetizes your activity on Meta platforms using your interactions (likes, shares, attention) to tailor your exposure to Meta’s massive advertising machine. So if you asked Meta AI about travel tips for your upcoming vacation, you might now see more content and ads related to that place. But what if youre asking Meta AI about how to deal with your depression?

In a blog post, the company shared:

“When people have conversations with Meta AI about topics such as their religious views, sexual orientation, political views, health, racial or ethnic origin, philosophical beliefs, or trade union membership, as always, we don’t use those topics to show them ads.”

But Meta has a spotty record when it comes to protecting sensitive personal information from leaking into its ad platform. Meta’s pixel-tracking technology has been found to pick up sensitive information regarding mental heath crises, financial information, and medical information.

Meta says you can manage the ads you see via controls in its privacy settings, but its unclear if users can opt out of the use of Meta AI conversations and interactions for ads and recommendations altogether.

The company said users will start to see notifications about the changes this month, which will go into effect on December 16, 2025.

Meta did not immediately respond to a request for comment.

Meta currently monetizes your activity on Meta platforms using your interactions (likes, shares, attention) to tailor your exposure to Meta’s massive advertising machine. So if you asked Meta AI about travel tips for your upcoming vacation, you might now see more content and ads related to that place. But what if youre asking Meta AI about how to deal with your depression?

In a blog post, the company shared:

“When people have conversations with Meta AI about topics such as their religious views, sexual orientation, political views, health, racial or ethnic origin, philosophical beliefs, or trade union membership, as always, we don’t use those topics to show them ads.”

But Meta has a spotty record when it comes to protecting sensitive personal information from leaking into its ad platform. Meta’s pixel-tracking technology has been found to pick up sensitive information regarding mental heath crises, financial information, and medical information.

Meta says you can manage the ads you see via controls in its privacy settings, but its unclear if users can opt out of the use of Meta AI conversations and interactions for ads and recommendations altogether.

The company said users will start to see notifications about the changes this month, which will go into effect on December 16, 2025.

Meta did not immediately respond to a request for comment.

tech

Tesla sales grow in some European countries for the first time this year

In September, Tesla sales rose in France and Denmark for the first time this year, while they also continued to grow in Norway and Spain, according to early European sales data reported by Reuters.

That’s a notable shift from the declines of previous months, as Tesla benefits from the rollout of its revamped Model Y and the introduction of numerous incentives across the continent, helping to stabilize its earlier sales slump.

Of course, Tesla’s European sales fluctuate dramatically month to month in Europe in part because the company doesn’t sell that many vehicles in Europe. Typically Tesla sells a few thousand vehicles per month per European country. In comparison, the automaker sells tens of thousands of cars in the US each month.

Tesla is up more than 2% in early trading, after having just capped off its best month since the election.

Stargate I in Abilene, Texas.

Rising ambitions and skyrocketing costs: Here’s what we know about Project Stargate

As the number of gigawatts and GPUs grows, so do the questions about how the massive data center project will be paid for.

tech

Apple reiterates plans to “partner with other generative AI chatbots” besides ChatGPT

Apple is playing the field with AI and it wants you to know.

In a filing to dismiss Tesla CEO Elon Musk’s lawsuit accusing the iPhone maker of favoring its partner OpenAI’s ChatGPT on the App Store, Apple said that can’t be the case because it is “widely known that Apple intends to partner with other generative AI chatbots.”

At its developer conference last year, Senior Vice President of Software Engineering Craig Federighi previously mentioned that Apple “intends to add support for other AI models in the future.”

Apple currently has a partnership with ChatGPT where users can direct their Siri queries to go through the chatbot. Apple, whose AI strategy has lagged its peers, has also been in talks with Anthropic and Google, and is reportedly considering using Gemini to power Siri.

Apple’s lawyers refuted X Corp.’s claims that Apple cannot partner with OpenAI “without simultaneously partnering with every other generative AI chatbot — regardless of quality, privacy or safety considerations, technical feasibility, stage of development, or commercial terms.” Apple’s legal team added, “Of course, the antitrust laws do not require that.”

Apple has yet to announce who its future AI partners will be.

At its developer conference last year, Senior Vice President of Software Engineering Craig Federighi previously mentioned that Apple “intends to add support for other AI models in the future.”

Apple currently has a partnership with ChatGPT where users can direct their Siri queries to go through the chatbot. Apple, whose AI strategy has lagged its peers, has also been in talks with Anthropic and Google, and is reportedly considering using Gemini to power Siri.

Apple’s lawyers refuted X Corp.’s claims that Apple cannot partner with OpenAI “without simultaneously partnering with every other generative AI chatbot — regardless of quality, privacy or safety considerations, technical feasibility, stage of development, or commercial terms.” Apple’s legal team added, “Of course, the antitrust laws do not require that.”

Apple has yet to announce who its future AI partners will be.

tech
Rani Molla

Meta buys chip startup Rivos in effort to lower its reliance on Nvidia

Meta is buying AI chip startup Rivos for an unknown sum, as part of the social media companys effort to decrease its reliance on graphics processing units from Nvidia, Bloomberg reports. Rivos was seeking funding in August at a $2 billion valuation. Meta has been spending exorbitant sums in an attempt to create AI models that are smarter than humans, an effort that’s involved investing in developing its own AI chips.

Latest Stories

Sherwood Media, LLC produces fresh and unique perspectives on topical financial news and is a fully owned subsidiary of Robinhood Markets, Inc., and any views expressed here do not necessarily reflect the views of any other Robinhood affiliate, including Robinhood Markets, Inc., Robinhood Financial LLC, Robinhood Securities, LLC, Robinhood Crypto, LLC, or Robinhood Money, LLC.