IBM launches z17: A personal AI supercomputer for enterprises

Hi Pals! Here’s what you need to know about AI today:

👉 IBM launches a new AI mainframe for enterprise customers

👉 Amazon’s Nova Reel can now generate videos of up to 2 minutes in length

👉 Google DeepMind is preventing its former employees to join competitors

and many more!

📧 Did someone forward you this email? Subscribe here for free to get the latest AI news everyday!

Read time: 5 minutes

What’s going on: IBM has launched the z17, its latest mainframe designed to meet the growing demands of AI adoption in enterprise settings. This system is powered by the IBM Telum II processor and tailored for over 250 AI use cases, including generative AI and AI agents. It targets industries like finance and retail, where mainframes remain critical, processing massive transaction volumes (up to 450 billion AI inferences daily), a 50% jump from the z16 released in 2022.

What does it mean: This computer aims to integrate AI directly into business operations, leveraging the mainframe’s secure, encrypted environment. The z17’s development spanned five years, predating the AI boom sparked by ChatGPT in late 2022. IBM incorporated feedback from over 100 customers, spending 2,000 research hours to align the system with market needs. The result is a mainframe that not only handles current AI workloads but is built with flexibility for future models. Energy efficiency is also improved over the z16, addressing practical concerns as AI power demands rise.

More details:

• Mainframes, often seen as legacy tech, are still relevant. 71% of Fortune 500 companies rely on them, and the market hit $5.3 billion in 2024.

• The z17 reinforces their relevance by embedding AI capabilities, allowing firms to process data locally rather than shifting to the cloud. This approach suits industries needing real-time insights, like fraud detection, while keeping infrastructure costs and security risks in check.

• The mainframe will support 48 IBM Spyre AI accelerator chips upon release, with the plan to bring that number up to 96 within 12 months.

• Interested in learning more about this mainframe’s details? Visit this page.

What’s going on: Amazon has upgraded its AI video model, Nova Reel, to generate videos up to two minutes long. This update builds on Nova Reel’s debut in December 2024, when it entered the competitive generative video market alongside models from OpenAI, Google, and Runway. The enhanced version, Nova Reel 1.1, can produce "multi-shot" videos with consistent styling across scenes.

What does it mean: This new update targets practical applications like marketing and content creation, where longer, cohesive videos are in demand. The model’s ability to maintain visual continuity over extended runtimes stems from improvements in its underlying architecture, though specifics remain undisclosed.

More details:

• Nova Reel competes in a crowded field, with rivals like OpenAI’s Sora and Google’s Veo 2, and recently released Runway’s Gen-4.

• Amazon’s focus appears to be on enterprise utility, integrating Nova Reel with its broader cloud services to appeal to businesses needing scalable, on-demand video generation.

• Amazon’s move to extend clip length addresses a practical limitation of earlier models, which often were limited to seconds-long outputs. While the market remains fragmented, Nova Reel’s two-minute capability positions it as a viable tool for users prioritizing efficiency over cinematic looks.

• Interested in the details and looking for tutorials of how to use it? Read Amazon’s official blog post.

🧠 Google DeepMind is reportedly using aggressive noncompete agreements in the UK to prevent AI employees from joining competitors for up to a year, paying them during this period.

📊 Meta's VP of generative AI, denied claims that the company artificially inflated the benchmark scores of its Llama 4 Maverick and Llama 4 Scout models by training them on test sets, attributing the reported "mixed quality" to the models being newly released and still undergoing implementation adjustments across different cloud providers.

🚗 Waymo is considering using interior camera data, linked to rider identities, to train generative AI models and potentially personalize ads, as revealed in an unreleased privacy policy.

🔍 Google's AI Mode, an experimental search feature, now allows users to ask complex questions about images using Google Lens' multimodal capabilities, understanding the scene, objects, and their relationships to provide detailed information and follow-up question support, and is now available to millions more users enrolled in Google Labs.

🛒 Shopify CEO has mandated that teams explore AI solutions before requesting additional headcount. It might potentially impact Shopify's workforce of approximately 8,100 employees.

📽 Moonvalley, an AI video startup developing AI tools for video creation, has raised $43 million in venture capital, and its Marey model offers customization options and is lower risk from a legal perspective.

AI + Automating tasks

Daisy - No-code & AI automation builder

Luma - AI Agent Engineer

Base - Senior Data Scientist

Tombras - Data Scientist

Thank you for staying with us like always! If you are not subscribed, subscribe here for free to get more of these emails in your inbox! Cheers!