Among funding announcements from new generative AI startups was underplayed AI/XR news from Meta, whose Ray Ban smart glasses got their expected boost this week with a Llama app upgrade. Apple is chewing its heart out with envy as it races to catch up. There’s plenty happening in the Metaverse, too, where Meta is not doing as well, spending gobs of cash on preteen-dominated Horizon Worlds in a seemingly futile attempt to catch up with Roblox and Fortnite. It is possible there is a game in tech Meta cannot play.

Veo 2, the current version, can generate 4K video clips from text prompts with remarkable control over cinematic elements like camera angles and lens types. It supports both text-to-video and image-to-video generation, and is now accessible via Gemini Advanced and Google AI Studio. There’s growing speculation that Google will announce Veo 3, alongside updates to its Gemini models, possibly including Gemini Ultra. Veo’s capabilities are also reaching consumers: the upcoming Honor 400 smartphones will feature a Veo-powered image-to-video tool that transforms static photos into five-second clips.

Founded in 2021, Hedra’s Character-3 foundation model blends text, images, and audio to generate lifelike digital avatars, aiming to help brands create animated mascots or virtual spokespersons. CEO Michael Lingelbach, a former stage actor, emphasizes the company's focus on overcoming the “uncanny valley” challenge in digital video performance. The funding round also included participation from a16z speedrun, Abstract, and Index Ventures. This round values the company at $200 million.

Founded by former OpenAI researcher Andrew Carr and Google Creative Lab’s Jonathan Jarvis, Cartwheel has launched with a $10 million funding round led by Craft Ventures. The company’s browser-based platform generates rigged character animations from text or video, reducing animation tasks from hours to minutes. It supports professional workflows with exports to Maya and Unreal, while remaining accessible to non-technical users. The team includes veterans from Pixar, Riot, and Unity, with early adopters from DreamWorks, Roblox, and Take-Two. Cartwheel emphasizes ethically sourced training data and aims to simplify animation for games, advertising, and education. An API is planned for later this year.

The company has developed a training platform that combines 3D reality capture, spatial AI, and contextual data to produce personalized, photorealistic simulations. Already in use at the U.S. Department of Defense, the system can create interactive training content 10 times faster and at a fraction of traditional costs. Users interact with the platform through an AI Instructor via voice or text, enabling real-time guidance and dynamic scenario adjustments. The round was led by Owl Ventures and joined by a16z speedrun, Alumni Ventures, and others.

AssetGen 2.0, an advanced AI model for generating detailed 3D assets directly from text prompts, is set to be integrated into Meta's Horizon Desktop Editor, to streamline the creation of immersive environments. It’s going to be so hard to catch up to Roblox and Fortnite creative. However, he is a Zuck of his word. The man has resources, conviction, and consistent vision. This can’t be the only reason they are doing this. Can it?

The app offers standard AI features like text and voice interaction, image generation, and web search. Its "Discover" feed allows users to view and interact with AI-generated content shared by friends on Instagram and Facebook, adding a social dimension to AI interactions. The app features an experimental “full-duplex” voice mode, providing more dynamic, conversational engagement. Currently available in the U.S., Canada, Australia, and New Zealand, the app utilizes Meta's Llama 4 model and personalizes responses using data from Facebook and Instagram. The app also replaces the former View app for Meta’s Ray-Ban smart glasses, integrating AI features like object recognition and real-time translation.

New ‘Commerce APIs,’ enable creators to sell physical merchandise directly within their games. Integrated with Shopify, users can now purchase real-world items like apparel and cosmetics inside the virtual world. Notably, Fenty Beauty is launching an exclusive lip gloss shade available only through its Roblox experience. Additionally, Roblox's Approved Merchandiser Program lets physical products include codes redeemable for digital avatar items. Early adopters have reported significant revenue through this integration.

The Avatar Marketplace enables users to buy avatars made by other users directly on the platform, using in-game currency redeemable for actual cash IRL. Users can now browse, try on, and purchase avatars directly with VRChat Credits. Creators’ cut is 50%. For the past five years, the VRChat community has been transacting on Fiverr, with creators charging over $1,000 for PC avatars equipped with features like a prehensile tail.

The great thing about being one of the biggest companies in the world is that no matter how badly you stumble, there is always another at-bat. Startups don’t have that luxury. Apple AI smart glasses are expected in 2026 or 202, with a display equipped model to follow soon thereafter. Apple hopes to dethrone Meta’s Ray Ban’s with a custom chip, based on the Apple Watch's low-power processors, managing multiple cameras which provide the real world context wearable AI needs.

This column serves as the script for the news segment of our weekly AI/XR Podcast, co-hosted by former Paramount futurist and co-founder of Red Camera, and Rony Abovitz, founder of Magic Leap, Mako Robotics, and Synthbee AI. This week’s guest is Ori Inbar, co-founder and CEO of AWE. You can find us on Spotify, iTunes, and YouTube.​