Larry Dignan

Walmart said it will acquire Vibe.co in a deal that expands its connected TV advertising reach. Walmart will combine Vibe.co's self-serve connected TV platform with Walmart's ad business, which includes VIZIO.

The Top500 list of the world's most powerful supercomputers has a new No. 1. China's LineShine, which is built on custom Chinese processors, has topped El Capitan. Five systems now cross the Exascale threshold.

Here's the list.

Key facts about the new No. 1.

• LineShine topped El Capitan for the top supercomputer.
• LineShine had a debut measurement of 2.198 Exaflop/s on the HPL benchmark, more than 20% ahead of the No. 2 system, using 13,789,440 cores.
• The system is based on the custom "LingKun" platform with 304-core LX2 processors running at 1.55 GHz, the proprietary LingQi interconnect, and Kylin OS.
  
  Read into this what you will, but pretty sure US angst is on deck. At the very least, China is showing it has custom silicon game.

Nvidia launched Nvidia Halos for Robotics, an open robotics stack that includes safety layers. The robotics version of Halos is an extension of Nvidia Halos autonomous vehicle safety systems. Nvidia's Halos for Robotics aims to provide a system that combines AI compute and safety for physical AI and robotics. Agility, a humanoid robotics company, is the first to use Nvidia Halos for Robotics in factories, warehouses and logistics.

The stack includes Nvidia IGX Thor and Nvidia Holoscan Sensor Bridge, Halos OS and an inspection lab. Nvidia detailed the effort in a news release and technical blog.

Getty Images said its licensed images and content libraries will be available in ChatGPT experiences and OpenAI Search. The Getty Images will be used to enhance the visual content in ChatGPT. Terms of the deal weren't disclosed.

Microsoft CEO Satya Nadella warned that AI supremacy can't be controlled by just a few companies with proprietary models. In an interview with The Wall Street Journal, Nadella said Microsoft is seeking to power a move away from frontier model concentration with low-cost models and task-built models.

Now this take is just swell given enterprises are being hit with AI sticker shock, but let's get real. Microsoft helped create OpenAI's momentum and counts Anthropic as a key partner. Now that Microsoft is creating its own models it's the champion of lowering the AI cost curve.

Nadella is simply talking Microsoft's book as traders would say. That said, Nadella isn't wrong. Enterprises need to avoid model lock-in and the smart CIOs are already using multiple models.

• Why Microsoft AI's approach is right time, right place
• Google Cloud, AWS, Microsoft Azure: The AI vertical integration race
• OpenAI goes multi-cloud in amended deal with Microsoft

Bloomberg's Mark Gurman has a riff on how incoming Apple CEO John Ternus will give the company's industrial design team more say. Ternus takes over for Tim Cook at a time when many argue that Apple's product lineup has stagnated.

Intel named former SK Hynix CEO Seok-Hee Lee as executive vice president of Intel Foundry. Lee will report directly to CEO Lip-Bu Tan. In a statement, Intel said Lee will "lead all advanced packaging, system integration, back-end technology development, and back-end manufacturing, strengthening Intel’s ability to deliver differentiated, system-level innovation for customers."

Tan said Intel Foundry is establishing an advanced packaging business because systems integration is critical to Intel Foundry growth.

Naga Chandrasekaran, executive vice president of Intel Foundry, will lead front-end technology development and front-end manufacturing and continue to report to Tan. Executive vice president Navid Shahriari will be retiring after 37 years at Intel.

Is it just me or is that a lot of EVPs for one press release?

Z.ai released GLM-5.2, its latest flagship model designed for long-horizon tasks and agentic coding with a 1M-token context. The model, which has an MIT open-source license, is the new leading open weights model on the Artificial Analysis Intelligence Index and leads MiniMax-M3, DeepSeek V4 Pro and Kimi K2.6.

More:

• Why enterprise AI leaders need to bank on open-source LLMs
• How to avoid vendor lock-in in the AI age
• AI inference costs are going to be a big concern: What's the fix?
• PCs and workstations get their AI inference moment
• Nvidia Nemotron: Much needed open-source model champion in US
• LLMs jockey for higher position as graders, reviewers, orchestrators

The Pragmatic Engineer has a good read on how Meta's engineering culture has taken a hit. The article is food for thought for those enterprises and leaders looking to overindex on AI.

"For two decades, Meta had a unique, high-performance engineering org; right up until around April of this year. For the first 20 years of the company’s existence, it had a “move-fast-and-break-things” culture, and in the early 2020s this shifted to a “move-fast-with-stable-infra” one. Engineers I know at the company were empowered to do good work, focus on impact, and to balance business interests with solid engineering.

But in the past few weeks, all that has changed, as if the leadership has been following detailed blueprints on how to demolish a proven, successful engineering culture in the most ruthlessly efficient way possible."

Midjourney, which is best known as an AI image generator, launched its hardware efforts with Midjourney Medical. The device, which will be part of a broader Midjourney spa effort, is a full-body ultrasound scanner. In a post that could be confused for a spoof if it weren't April Fool's, reads in part:

"It starts by stepping into a shallow pool of golden light. You then begin to descend into the water. Your body passes through a ring of underwater sensors, each acting like a dolphin, using its echolocation. The sensors send ultrasonic sound waves through your body from every angle. With enough waves, and enough angles, we form an image of what's happening inside your body.

The goal is for this process to take no more than 60 seconds.

You go into the water, you come out of the water, and you're done."

I'm probably done thinking about this one until I see one of these spas in San Francisco.

Midjourney said:

"The next 12 months are about refining our algorithms and hardware on a daily basis. We’ll be doing research trials to show off the raw capabilities of our system, moving towards a 2nd generation hardware design, and we’ll do build-out for our first “research spa” which will become the promethean site that enables mass-scale health scanning.

Around the end of 2027 we’ll open up the Spa, and we’ll begin getting real world knowledge of what this infrastructure is going to be like."

Noam Shazeer, who invented the transformer and mixture of experts scaling architectures for large language models, is leaving Google Deepmind for OpenAI. Shazeer announced the move on X. Shazeer had been at Google for most of his career in two stints. He left Google to be CEO at Character.ai and then was acquired by Google for $2.7 billion. Since August 2024, Shazeer was VP of Engineering and Gemini Co-Lead.

South Korean content generated 12.1 billion hours of Netflix viewing globally from April 2025 to March 2026 and ranked second only to the US among content origins on Netflix, according to Omdia.