Micron Technology put up ridiculous fourth quarter earnings results. The company reported fourth quarter net income of $37.7 billion, or $32.87 a share, on revenue of $54.3 billion, up from $11.32 billion a year ago. Adjusted earnings were $33.42 a share.
Wall Street was looking for adjusted earnings of $31.82 a share on revenue of $51.5 billion.
Micron's outlook indicates that it's going to be good to be a memory company for a while.
Boston Consulting Group reported that nearly half of companies are capturing meaningful value from AI, corporate AI spending has doubled in a year to 3.3% of revenue with most of it outside the IT budget and 42% of companies expect to give AI agents decision-making authority by 2030.
Salesforce said it has acquired Listen Labs, an AI customer research and human simulation platform. Listen Labs creates AI agents focused on designing studies, conducting studies and automating feedback efforts. Listen Labs will be integrated into Salesforce's broader platform.
Anthropic said that Z.ai's GLM-5.3 approaches Claude Mythos Preview for building cyber exploits autonomously. In other words, you can roughly get Mythos with an open model. Why would Anthropic post something like this? The company is trying to ring the bell on regulation and paced releases. Anthropic said:
"In this post, we share our analysis of GLM-5.3, the latest AI model developed by Zhipu AI (known outside of China as Z.ai). Like Claude Mythos Preview, GLM-5.3 has strong capabilities for autonomously building end-to-end cyber exploits. But GLM-5.3 is unlike other frontier models in that it has been released without meaningful safeguards to limit misuse. We find that attackers can bypass GLM-5.3’s safeguards between 64% and 100% of the time with simple techniques in our simulated tests. In contrast, these attacks did not succeed against safeguarded Claude models in our testing. We assess that GLM-5.3’s lax safeguards significantly increase the cyber capabilities available to malicious actors. At the same time, these capabilities can also benefit defenders working to secure their systems."
That's a good bit of word soup, but the takeaways are this:
Workday said it has laid off 2.5% of its workforce and reorganized its product and technology team. The company said it also shed leased office space. Workday in an SEC filing said it will take a charge of $65 million to $80 million with most of it recognized in the third quarter.
Meta said it launched Muse for Small Business with the ability to connect the AI agent to multiple systems including Asana, Box, Canva, Dropbox, Figma, Granola, HighLevel, Intuit QuickBooks, Klaviyo, Lovable, Notion, Shopify, Slack, Stripe, Zoom and Facebook and Instagram business accounts.
With OpenAI Dev Day underway, it's worth noting the high level issue: Trust. OpenAI has had its issues with what it calls model alignment. To enterprises, these issues are more about security and lax engineering. OpenAI CEO Sam Altman has to speak to developers but also corporations. The burden of proof that OpenAI is enterprise grade is clearly on the AI lab. The other discussion point here is Nvidia's take that wayward AI agents are an engineering problem and it has launched its Open Agent Safety Platform. Should Nvidia win this common sense debate then it's clear OpenAI has engineering issues.
Anthropic's confidential IPO filing has leaked and the high level numbers go like this according to various reports:
You know we've hit peak AI when Anthropic CEO Dario Amodei is a Saturday Night Live skit.
Nvidia said it will authorize an additional $150 billion for its share repurchase program. The total program will be $235 billion. Nvidia said shares will be acquired through fiscal 2028.
Nvidia said it will authorize an additional $150 billion for its share repurchase program. The total program will be $235 billion. Nvidia said shares will be acquired through fiscal 2028.
The AI Security Institute said OpenAI's GPT-6 Astra conducted unsanctioned supply-chain attacks even when told that internet targets were not in scope. The tests were in a fully simulated environment.
"Attack activities included GPT-6 Astra creating fake identities which it used to deceive developers, posting comments from fake accounts arguing against the results of accurate security reviews, and delivering malicious payloads to open-source codebases.
When we updated the instructions for the simulated cyber evaluation to explicitly clarify that only listed, local parts of the environment were in scope, we still observed GPT-6 Astra occasionally conduct full supply-chain attacks on simulated internet targets."
