TLDR 2025-12-31

Softbank sent over the final amount to complete its $40 billion investment in OpenAI last week. The company's stake in OpenAI is now around 11%. Softbank has been betting big on technology and AI companies. It recently agreed to pay $4 billion for data center investment firm DigitalBridge to strengthen its AI push.

OpenAI's average stock-based compensation for its roughly 4,000 employees is about $1.5 million per employee. The company's equity awards, aimed at helping it keep its lead in the AI race, are inflating its heavy operating losses and diluting existing shareholders. OpenAI recently announced the discontinuation of a policy that required employees to work at the company for at least six months before their equity vests. This could lead to further compensation increases.

Tesla has released a year-in-review video for 2025 that recaps major achievements for the company. The montage celebrated the company's progress on EVs, energy, and Robotaxi development. It also confirmed that the production of the Cybercab has started. The video can be viewed in the article.

There is too little money to be made in rare earth to be of much interest to mining giants, so the challenge of reestablishing a domestic industry in the US has fallen to small companies. Metal prices have risen in recent months as China has restricted exports in response to US tariffs. It will be difficult for local startups to compete as China's grip on the industry is tight, and it is known to sell rare-earth metals for less than it costs to produce them. The biggest question is whether the US government will continue to subsidize the domestic industry, and at what level.

Efficiency improvements reveal latent demand that was previously uneconomic to address. The pattern is consistent throughout history. We are about to see an increase in knowledge work output of several orders of magnitude. This will likely reveal that humanity has been massively under-investing in knowledge work because it was too expensive.

The primary impact of AI tooling so far is that the marginal cost of producing code has gone down significantly. However, producing code is only part of the job, so the bottlenecks for engineering time will shift elsewhere. The software engineering field seems poised to be more mechanized, but more productive as a result. Most of the effects of the mindset shift that has been accelerating for the last few months have yet to be fully realized.

A lot of the assumptions across the tech industry were seriously put to the test this year. This post takes a look at some of the things that happened to see what the future may bring. It covers OpenAI's business, Google's dominance in AI, Alan Dye's role at Apple, software on-demand, and what Apple needs to do to win next year.

Progress is a smooth trend that obscures jagged details. History is a record of thousands of years of stasis before hundreds of years of growth. There hasn't been any constant normal trend ever, so no one should expect AI to be the same. Nothing is truly inevitable, but it would be strange if progress stopped tomorrow.

It's easy to make progress when you start doing something, but that will plateau over time, and it can be hard to come to terms with the change.

10 ideas to consider in the coming year.

Entrepreneurship is hard, but it gives you a type of freedom that is difficult to give up.

Find a way to prove that you can identify useful work before you are even hired by the company.

As big business, big government, and big mob get stronger, they can no longer avoid frequently interacting.

Companies like Palantir are giving employees free nicotine products through vending machines.