AI's Latest Buzz: Mistral's Mega Funding, Apple's 'Superpower' Cameras, and India's Culturally Aware Video AI
Catch up on the latest AI news: Mistral AI's potential €20B valuation, Apple's measured approach to AI in iOS 27 cameras, Avataar AI's culturally aware video model for India, Meta Edits' new AI assistant, and Apple's cautious stance on agentic AI.
The world of AI and tech is buzzing with significant developments today, from massive funding rounds for European AI pioneers to innovative applications tailored for unique markets and cautious approaches to generative capabilities. We're seeing AI evolve from abstract concepts into tangible tools empowering creators, enhancing photography, and driving competitive strategy across continents.
TL;DR
- Mistral AI is reportedly in talks to raise €3 billion, potentially valuing the company at €20 billion, solidifying its position as a European AI leader.
- Avataar AI launched Varya, a video generation model optimized for India's diverse cultural context, making video creation 10 times faster and cheaper.
- Apple's Camera Chief, Jon McCormack, highlights that iOS 27's new AI features, like Extend and Spatial Reframe, will give users "superpowers" in photography without "AI for the sake of AI."
- Meta's Edits app is rolling out a new AI assistant and a desktop version, aiming to boost creator engagement and streamline content insights.
- Apple is taking a deliberate and less hyped approach to agentic AI compared to its peers, focusing on practical user utility in its Siri AI and Photos app.
Mistral is rumored to be raising €3B at €20B valuation
French AI lab Mistral AI is reportedly in early discussions to secure approximately €3 billion ($3.5 billion) in funding. This significant round would project the company's valuation to reach around €20 billion (about $23.15 billion), a substantial leap from its previous €11.7 billion valuation just last September.
Mistral AI, established in 2023, has quickly emerged as a frontrunner among Europe’s AI startups. The company is distinguished by its more open stance on AI development, providing some foundational large language models with open weights, allowing for broader customization. Alongside these open models, Mistral also offers specialized closed models for applications like programming, voice cloning, and optical character recognition.
In a strategic move to cater to Europe's increasing desire for technological independence, Mistral has positioned itself as a "sovereign" and homegrown alternative to American tech giants. The company is actively setting up a data center near Paris and has forged partnerships to bolster its European presence.
The potential €20 billion valuation underscores Mistral AI's rapid ascent and its strategic importance in the global AI landscape, particularly within Europe.
Cheaper, faster, and culturally aware, Avataar’s video AI is built for India’s scale
India's AI development, while lagging behind the U.S., Europe, and China, is receiving a significant boost from initiatives like the India AI Mission, a $1.2 billion government program. This initiative aims to spur innovation by providing subsidized GPU compute access to selected startups in exchange for public model releases. Among the 12 startups chosen, Avataar AI has launched a groundbreaking video model named Varya.
Varya is specifically designed with a deep understanding of local context, capable of identifying unique Indian festivals, foods, and clothing. Instead of building from scratch, Avataar AI, backed by Peak XV, utilized a technique called distillation, starting with Alibaba's publicly available Wan 2.2 model. This process compressed Wan 2.2's capabilities into a leaner, faster version, optimizing it for Avataar's e-commerce video tools.
The result is a highly efficient model that executes in just four steps compared to Wan 2.2's 50, enabling Varya to generate video 10 times faster and at a fraction of the cost. For example, using an NVIDIA H200 GPU, Varya can produce a 5-second 720p clip in 45 seconds, a dramatic improvement over the 1,230 seconds it would otherwise take.
Avataar AI's Varya model represents a significant leap in localized AI, making culturally relevant video generation accessible and efficient for the Indian market at an unprecedented scale.
Apple’s Camera Chief Thinks AI Can Give You Superpowers
Apple is introducing new generative AI features to its Photos app in iOS 27, set to arrive on iPhones later this year. While other tech giants like Google and Samsung have already integrated advanced generative AI for photo manipulation, Apple's iPhone camera chief, Jon McCormack, emphasizes a more deliberate approach, stating they are not "doing AI for the sake of AI."
At its annual Worldwide Developer Conference (WWDC 2026), Apple showcased features like an enhanced Clean Up tool for erasing unwanted objects. More notably, new features called Extend and Spatial Reframe will allow users to expand photo backgrounds or alter image perspectives by generating "fake pixels." This capability enables the camera to intelligently infer and draw in what should be there, empowering users with advanced editing previously requiring professional software.
McCormack highlights that AI is helping to solve a "giant backlog of unsolvable issues," giving "normal people these absolute superpowers" without needing expert knowledge in tools like Photoshop. Apple's strategy focuses on practical, deliberate AI applications within the Photos app, though the article notes that more radical generative AI capabilities might be available through the App Store.
Apple's new iOS 27 AI features in the Photos app aim to democratize advanced image manipulation, providing users with "superpowers" through intelligent, generative pixel creation rather than simply adding AI as a trend.
Meta’s Edits app is getting an AI assistant and a desktop version
Meta recently previewed significant updates to its video editing app, Edits, at an invite-only creator event in L.A. These upcoming additions include a new AI assistant and a much-anticipated desktop version, moving beyond its initial mobile-only format. Launched last year as a direct rival to ByteDance’s CapCut, Meta is enhancing Edits to retain and attract more users in the competitive short-form video landscape.
The forthcoming AI assistant is designed to provide creators with valuable insights and content ideas by analyzing their Instagram data, such as views and video retention rates. It will help creators understand what content resonates with their audience and suggest new video concepts, including leveraging trending audio. This integration aims to streamline the creative process, reducing the need for creators to use external tools for analysis and brainstorming.
By offering an AI assistant directly within Edits and expanding to a desktop platform, Meta seeks to keep creators engaged within its ecosystem, thereby encouraging more frequent posting and boosting overall user engagement. This move is part of Meta's broader strategy to compete effectively with platforms like TikTok and YouTube for creator attention and content.
Meta's integration of an AI assistant and a desktop version into its Edits app is a strategic move to empower creators with data-driven content ideas and a more flexible workflow, strengthening its position against rivals like CapCut and TikTok.
I'm Glad Apple Isn't Hyping Up Agentic AI (Yet)
While "agentic AI" has been the latest buzzword dominating recent tech keynotes from Google I/O, Microsoft Build, and NVIDIA's Computex, Apple has taken a decidedly different and less hyped approach. Agentic AI refers to intelligent agents capable of performing tasks autonomously, such as scheduling meetings from emails. However, concerns about the reliability of current AI models, which can still "hallucinate" or be untrustworthy, raise questions about the wisdom of fully autonomous agents at this stage.
During its WWDC 2026 keynote, Apple only briefly touched upon agentic AI. Instead, the company focused extensively on demonstrating how its new Siri AI could offer practical utility to users. Examples included finding specific information within lengthy text message threads or assisting with securing tickets for exclusive concerts. Siri AI primarily responds to user commands, but leverages modern AI models for improved data synthesis and responsiveness.
Early observations of Siri AI at WWDC 2026 suggest it performs as advertised, though long-term testing will be crucial to assess its sustained reliability. This measured approach, combined with Apple's commitment to privacy through Private Cloud Compute, positions it distinctly from competitors. The article notes an intrigue in Siri AI's possibilities, even for an AI skeptic, more so than for other offerings like Microsoft's Copilot.
Apple's pragmatic focus on making Siri AI genuinely useful through enhanced data synthesis and user-commanded actions, rather than aggressively hyping agentic AI, sets it apart from industry trends and addresses concerns about AI reliability.