Meta has released Muse Wonder and Muse Wonder Full in what it calls its most advanced models yet for personal and professional agentic AI tasks. Muse Wonder is available for developers and Muse Wonder Full is available for Muse Code subscribers. Muse Wonder will be the default model in Muse shortly.
Intel CEO Lip-Bu Tan and Broadcom CEO Hock Tan both said that capital could become the main constraint to the AI buildout. According to CNBC, Tan said capital is could become restrained due to interest rates and demand for returns. Go figure.
Amazon is reportedly in talks to buy Decart for $7 billion, according to the Wall Street Journal. Decart makes it easier to switch AI workloads between different processors. Combined with model routing, you can play this out where you'll route models as well as GPUs to optimize for cost. Given AWS has multiple instances and processors such as Trainium, Decart could be a good fit.
Microsoft CEO Satya Nadella talks about the need for deterministic controls for agentic AI systems and the conversation quickly turns to the great super intelligence sellout. Meanwhile, Marc Benioff said Agentforce will become SIForce on X. Note that Salesforce's web site hasn't changed so it's unclear whether the branding team knows yet. That's probably good because you could easily see SIForce as systems integration force. But I digress. My strategy with "super intelligence" goes like this:
Zscaler CEO Jay Chaudhry said during the company's investor day:
"The word platform has lost meaning. Every company claims to do a platform. Platform actually uses a common set of services for policies. That's a real platform. I see many customers talk about and say they got sold a platform. When they looked underneath, they found that it is 8 policy engines, 8 products.”
Fun fact: The word “platform” was said 113 times during Zscaler’s investor day.
Global PC shipments fell 20.1% in the third quarter due to supply chain constraints and inventory pull-in as vendors stocked up before memory prices surged, according to IDC. In addition, high PC prices have hurt demand. IDC noted there may be promotions, but relief for PC buyers will be limited and "don't expect pricing anywhere near where it was a year ago." Here's a look at the standings:
Anthropic updated some of its usage rules and the most notable one is that you better be nice to Claude. According to updated terms:
"We’ve added a prohibition on sustained and needless abusive or cruel behavior toward our models. The policy update is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose. It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research.
This addition aligns with a step we've already taken, allowing Claude models to end rare conversations with persistently abusive users on Claude.ai and Claude Code. Such abuse is the main focus of this update; Claude’s ability to end these interactions will remain the primary enforcement mechanism."
It's pretty clear that Anthropic sees Claude as a person. Yay anthropomorphism.
The Association for Human Mathematics has urged its members not to work with OpenAI after the company released more than 700 files on GitHub showing an unreleased model's mathematics work. Meanwhile, computer science and quantum computing researcher Scott Aaronson said frontier model companies, notably OpenAI, are exploring whether their models can break cryptography. He pointed to a more measured statement from the Advisory Group on Mathematics and Artificial Intelligence.
HPE announced a set of HPE ProLiant Gen13 Servers powered by AMD's 6th generation AMD EPYC processors. The next-gen HPE ProLiant DL525 and HPE ProLiant DL585a servers have higher core density, memory bandwidth and I/O. HPE also launched two servers, HPE ProLiant XD245 and HPE ProLiant XD285, built on Open Compute Project standards for high rack density deployments.
Microsoft opened preorders for the Surface Laptop Ultra starting at $2,599. The laptop features and RTX Spark chip. Microsoft also opened preorders for the $5,999 Surface RTX Spark Dev Box, which has 128GB of unified memory to run models with more than 120B parameters.
CoreWeave said it is working with AdaniConneX to build 240 MW of capacity in India. CoreWeave will be sole tenant and will deploy Nvidia's Vera Rubin platform in the facility.
Fully Connected 2026 coverage:
CoreWeave customer stories:
Automation Anywhere said it will acquire Boost.ai from Nordic Capital. The move gives Automation Anywhere voice AI tools for customer experience. The Boost.ai deal complements the company's acquisition of Aisera, which added enterprise knowledge and employee experience tools to Automation Anywhere's platform.
