The Morning Download: It’s an AI Evolution

What's up: U.S. clamps down on investment in Chinese tech companies; Meta is developing a new AI model code-named ‘Mango’; OpenAI unveils an updated version of GPT-5.2

Good morning. AI remains omnipresent in the final days of this year. From OpenAI’s latest fundraising plans to Meta’s image generator and investments in China, the AI narrative is expanding faster than the Taylor Sheridan omniverse. (Much more on that below, including highlights of my recent conversation with Atlassian co-founder and CEO Mike Cannon-Brookes.)

The power of the AI story has been tested this year in the financial markets, and that’s a good thing. The role of the market is to root out excess and weakness, and there’s no doubt that process will continue with respect to AI. But I think the fundamentals of the AI story are intact. Demand remains strong, among businesses and consumers alike. Companies, led by their CEOs, continue to invest in AI.

While wholesale corporate transformation is elusive, many companies tell us that AI is worth the investment in the situations where it has been put to use. These steady gains and benefits, often measured by the recapture of time, are bound to continue. The AI revolution is perhaps better understood as an evolution, and I think that’s a more sustainable approach for companies at the forefront of AI adoption.

I think AI adoption in the enterprise will continue to evolve in this manner in the coming year. I know, many studies and research papers suggest that most corporate AI initiatives fail to deliver a significant return. That may be true, but most of these studies miss a crucial point. The organizations that do drive a reasonable return on AI are among the biggest and most influential companies in the world, and they are getting stronger. Sure, Walmart is just one company, but it’s a significant and influential part of the economy.

I will be taking a break from work for the next two weeks. I wish all of you happy holidays and the very best in the coming year. Thanks for reading and supporting CIO Journal, the Morning Download, and the WSJ Leadership Institute, which is off to an amazing and exciting start. And thanks for sharing your ideas and insights.

Is AI going to kill software? No, it’s not, Atlassian co-founder and CEO Mike Cannon-Brookes told me during a recent conversation at the company’s offices in New York. He was in town to ring the bell at the Nasdaq, where the Australian company marked its 10th anniversary as a public entity. Here are edited highlights:

WSJLI: How is AI reshaping software and your products, specifically Jira, Confluence and Loom?

Cannon-Brookes: SaaS is a technology delivery mechanism, AI is a technology disruption. They are sort of...almost orthogonal actually to each other, which is where I think this is a bit of a challenge. AI was honestly one of the best things that ever happened for us, for our mission. We have a huge amount of people problems to go solve. We have a massive customer base...and you give us this great new technology piece that arrived from the heavens...and we have the R&D resources to adapt and build and make that fit for customer problems. I think the difference is we are not trying to sell AI products; we want to use AI to make [products] better.

WSJLI: What is the impact of AI on workflow?

Cannon-Brookes:  They're grains of sand in the gears of every collaborative loop in these businesses. And every time you get the grains out, [it] just runs a little smoother. Just a pretty common thing. I can tell an internal example from our finance team. We have one person in finance whose job it is to jump in, to answer travel expense questions. 'Can I spend $200 on lunch?'...This person was answering tens of these questions every week. They made themselves an [AI] engine just so that they didn't have to do this part of the job. The agent just basically tried to supply the person with the answer as soon as they asked a question. Pretty simple...It goes to the AI use case.

WSJLI: How is AI shaping leadership?

Cannon-Brookes: Inside the company, we talk about applied AI. In applied AI, we talk a lot about 'AI Joy,' and what we mean is fundamentally, I want to use all of these technologies to make your job more joyful, to have your outputs be high quality. I want to get rid of as many of the [tedious] processes as possible so you have a bit of time. I want people to fail [at AI] quite a bit. Right now what we need them to do is to understand the technology in a really deep, intrinsic way. I feel like the only way to do that is to have them play with it...The best way to get 4,000 people 'AI ready'...is just to use it in all different forms.

WSJLI: Is AI destroying jobs on a mass scale?

Cannon-Brookes: You should be scared about somebody who's really good at using AI taking your job. I don't think these jobs [will be] replaced by machines. I just literally do not, at the macro, large-scale level, believe that's what happened. What's more likely to happen is an employee who doesn't understand these technologies is not getting the same quality of outputs [as] someone that does understand these technologies... that's why we want to promote learning. We want all of those groups to have an understanding of how AI can interact with them.

Image and video generation is the new front line in the AI arms race. Google’s Nano Banana image tool helped catapult Gemini into the spotlight earlier this year. And now Meta and its newly assembled AI dream team is aiming to strike back.

Meta Platforms is developing a new image- and video-focused AI model code-named Mango, alongside a next-generation text model called Avocado. Meta’s chief AI officer, Alexandr Wang, talked about the models, as well as work exploring so-called world models, in an internal meeting on Thursday, WSJ reports. Mango and Advocado are expected to be released in the first half of 2026.

OpenAI on Thursday unveiled an updated version of its flagship GPT-5.2 model for Codex, its AI-based coding agent. The company said the model is built on GPT-5.2’s strengths in professional knowledge work and GPT-5.1-Codex-Max’s agentic coding and terminal-using capabilities. Businesses including Cisco, Ramp, Virgin Atlantic and Gap are using the Codex agent, according to OpenAI.

OpenAI is also aiming to raise as much as $100 billion. The fundraising round, which is in the early stages, could value the company at as much as $830 billion, WSJ reports.

Anthropic said Wednesday it would release its Agent Skills technology as an open standard, a strategic bet that sharing its approach to making AI assistants more capable will cement the company's position in the fast-evolving enterprise software market, VentureBeat reports. The company also unveiled organization-wide management tools for enterprise customers and a directory of partner-built skills from companies including Atlassian, Figma, Canva, Stripe, Notion and Zapier.

French AI startup Mistral unveiled a new model this week, called OCR 3, that helps companies extract text and images from paper documents and PDFs, VentureBeat reports. In doing so, Mistral is betting that enterprises will need to digitize their documents in order to start getting value out of AI.