In previous memos I've written on CommonStock, I've been bullish on SF real estate and on office real estate in general because of two reasons: (1) office real estate today looks unfavorable today like retail real estate was during the time when people thought $AMZN would put all other retail businesses in bankruptcy and (2) the most innovative companies in the world, like Waymo, Anthropic, and OpenAI, have been capitalizing on the discounts in office leases in the city. For Waymo, the ROI is easy to understand since they are the sole robotaxi operator in the city and that's where they've done majority of their testing. For Anthropic and OpenAI, being in the tech sector, one would assume they'd be open to remote work like their tech peers when in reality, they are against remote work.

Unlike most work in the tech sector, the work being done on AI requires collaboration. Having AI researchers and other professionals in the field work in-person at a secure environment like the office helps these startups create ground-breaking AI applications at an accelerated pace. If these startups chose to operate 100% remotely, the financial savings that they would reap by not leasing and maintaining an office space would be far less than the value they could've created if they brought their workers to the office.

In this memo, we will analyze the financial benefits that OpenAI gets from signing the biggest office lease in SF since 2018. This building is the former $UBER HQ.

Here are the financial assumptions:

These assumptions are based on actual sources. It will be key to understanding the minimum amount of value that the employees in this office need to create to justify the lease.

Here are the assumed output benefits that would come with bringing employees to the office:

These are estimates that I've created and are not estimates that OpenAI nor anyone else is forecasting. Everything here is theoretical and there is no research paper, data, or any other source that I used to make up nor could use to justify these numbers.

When including the productivity benefits that comes with bringing employees to the office, on average, each employee will contribute an additional $101,250 in value to OpenAI [($405,000,000 / 4,000) = $101,250]. The value per employee increases to $1,828,004.11 (revenue per employee + estimated annual benefits of bringing employees to the office per employee).

From a broader perspective, OpenAI's decision to sign the office lease is a sound one. The lease will generate significantly more value for the company than the cost of the lease itself. Even if we conservatively estimate the annual benefits at $100 million (after removing the theoretical benefits related to reduced hiring costs and increased investment), the ROI of the lease is still over 330%. This means that OpenAI will generate over three times as much value from the lease as it will spend on the lease itself. In addition to the financial benefits, the lease will also provide OpenAI with a number of other benefits, such as improved collaboration and productivity, the ability to attract and retain top talent, and an enhanced brand image.

The risk of impairment on the lease is also low. OpenAI has a strong track record of innovation and the nature of AI research favors in-person collaboration over remote collaboration. The move to capitalize on the cheap office rents in the city is backed with a strong purpose and the contrarian move by OpenAI can be considered rational after going over the analysis of the office lease benefits.