Latest news with #Hedra
CNBC
02-07-2025
- Entertainment
- CNBC
AI virtual personality YouTubers, or ‘VTubers,' are earning millions
One of the most popular gaming YouTubers is named Bloo, and has bright blue wavy hair and dark blue eyes. But he isn't a human — he's a fully virtual personality powered by artificial intelligence. "I'm here to keep my millions of viewers worldwide entertained and coming back for more," said Bloo in an interview with CNBC. "I'm all about good vibes and engaging content. I'm built by humans, but boosted by AI." Bloo is a virtual YouTuber, or VTuber, who has built a massive following of 2.5 million subscribers and more than 700 million views through videos of him playing popular games like Grand Theft Auto, Roblox and Minecraft. VTubers first gained traction in Japan in the 2010s. Now, advances in AI are making it easier than ever to create VTubers, fueling a new wave of virtual creators on YouTube. The virtual character – whose bright colors and 3D physique look like something out of a Pixar film or the video game Fortnite – was created by Jordi van den Bussche, a long time YouTuber also known as kwebbelkop. Van den Bussche created Bloo after finding himself unable to keep up with the demands of content creation. The work no longer matched the output. "Turns out, the flaw in this equation is the human, so we need to somehow remove the human," said van den Bussche, a 29-year old from Amsterdam, in an interview. "The only logical way was to replace the human with either a photorealistic person or a cartoon. The VTuber was the only option, and that's where Bloo came from." Bloo has already generated more than seven figures in revenue, according to van den Bussche. Many VTubers like Bloo are "puppeteered," meaning a human controls the character's voice and movements in real time using motion capture or face-tracking technology. Everything else, from video thumbnails to voice dubbing in other languages, is handled by AI technology from ElevenLabs, OpenAI's ChatGPT, Google's Gemini and Anthropic's Claude. Van den Bussche's long-term goal is for Bloo's entire personality and content creation process to be run by AI. Van den Bussche has already tested fully AI-generated videos on Bloo's channel, but says the results have not yet been promising. The content doesn't perform as well because the AI still lacks the intuition and creative instincts of a human, he said. "When AI can do it better, faster or cheaper than humans, that's when we'll start using it permanently," van den Bussche said. The technology might not be far away. Startup Hedra offers a product that uses AI technology to generate videos that are up to five minutes long. It raised $32 million in a funding round in May led by Andreessen Horowitz's Infrastructure fund. Hedra's product, Character-3, allows users to create AI-generated characters for videos and can add dialogue and other characteristics. CEO Michael Lingelbach told CNBC Hedra is working on a product that will allow users to create self-sustaining, fully-automated characters. "We're doing a lot of research accelerating models like Character-3 to real time, and that's going to be a really good fit for VTubers," Lingelbach said. Character-3's technology is already being used by a growing number of creators who are experimenting with new formats, and many of their projects are going viral. One of those is comedian Jon Lajoie's Talking Baby Podcast, which features a hyper-realistic animated baby talking into a microphone. Another is Milla Sofia, a virtual singer and artist whose AI-generated music videos attract thousands of views. These creators are using Character-3 to produce content that stands out on social media, helping them reach wide audiences without the cost and complexity of traditional production. AI-generated video is a rapidly evolving technology that is reshaping how content is made and shared online, making it easier than ever to produce high-quality video without cameras, actors or editing software. In May, Google announced Veo 3, a tool that creates AI-generated videos with audio. Google said it uses a subset of YouTube content to train Veo 3, CNBC reported in June. While many creators said they were unaware of the training, experts said it has the potential to create an intellectual property crisis on the platform. Creators are increasingly finding profitable ways to capitalize on the generative AI technology ushered in by the launch of OpenAI's ChatGPT in late 2022. One growing trend is the rise of faceless AI channels. These are run by creators who use these tools to produce videos with artificially generated images and voiceover that can sometimes earn thousands of dollars a month without them ever appearing on camera. "My goal is to scale up to 50 channels, though it's getting harder because of how YouTube handles new channels and trust scores," said GoldenHand, a Spain-based creator who declined to share his real name. Working with a small team, GoldenHand said he publishes up to 80 videos per day across his network of channels. Some maintain a steady few thousand views per video while others might suddenly go viral and rack up millions of views, mostly to an audience of those over the age of 65. GoldenHand said his content is audio-driven storytelling. He describes his YouTube videos as audiobooks that are paired with AI-generated images and subtitles. Everything after the initial idea is created entirely by AI. He recently launched a new platform, TubeChef, which gives creators access to his system to automatically generate faceless AI videos starting at $18 a month. "People think using AI means you're less creative, but I feel more creative than ever," he said. "Coming up with 60 to 80 viral video ideas a day is no joke. The ideation is where all the effort goes now." As AI-generated content becomes more common online, concerns about its impact are growing. Some users worry about the spread of misinformation, especially as it becomes easier to generate convincing but entirely AI-fabricated videos. "Even if the content is informative and someone might find it entertaining or useful, I feel we are moving into a time where ... you do not have a way to understand what is human made and what is not," said Henry Ajder, founder of Latent Space Advisory, which helps business navigate the AI landscape. Others are frustrated by the sheer volume of low-effort, AI content flooding their feeds. This kind of material is often referred to as "AI slop," low-quality, randomly generated content made using artificial intelligence. "The age of slop is inevitable," said Ajder, who is also an AI policy advisor at Meta, which owns Facebook and Instagram. "I'm not sure what we do about it." While it's not new, the surge in this type of content has led to growing criticism from users who say it's harder to find meaningful or original material, particularly on apps like TikTok, YouTube and Instagram. "I am actually so tired of AI slop," said one user on X. "AI images are everywhere now. There is no creativity and no effort in anything relating to art, video, or writing when using AI. It's disappointing." However, the creators of this AI content tell CNBC that it comes down to supply and demand. As the AI-generated content continues to get clicks, there's no reason to stop creating more of it, said Noah Morris, a creator with 18 faceless YouTube channels. Some argue that AI videos still have inherent artistic value, and though it's become much easier to create, slop-like content has always existed on the internet, Lingelbach said. "There's never been a barrier to people making uninteresting content," he said. "Now there's just more opportunity to create different kinds of uninteresting content, but also more kinds of really interesting content too."
Fast Company
30-06-2025
- Business
- Fast Company
Startups are ditching LinkedIn for TikTok to announce funding rounds
The classic funding announcement post is getting the Gen Z treatment. More startups, especially those led by young founders, are moving away from LinkedIn posts or X threads and turning to viral TikToks and short-form videos to stand out, Business Insider recently reported. With traditional media coverage harder to land and social posts quickly vanishing from feeds, founders are rethinking how they announce major milestones. Cluely, the 'cheat on everything startup,' recently raised $15 million and announced it with a shot-for-shot homage to The Social Network. Earlier this year, they also launched with a video that cost $140,000 to produce. The 90-second narrative short, posted to X, shows a man on a first date being fed lines in real time by Cluely. The investment paid off. The video went viral and crashed Cluely's servers, founder Chungin 'Roy' Lee told Business Insider. For Hedra, a startup focused on digital avatars, the announcement video doubled as a product demo. Founder Michael Lingelbach appears in the clip as himself, as a Studio Ghibli character, a Pixar animation, and with a full tech bro makeover, complete with gold chain. Those viral baby podcast videos that were everywhere last month? That was them. Not every announcement needs to be a high-budget production, though. British entrepreneur Grace Beverley turned to TikTok last year to announce two fundraises: one for her activewear brand TALA, and another for Retrograde, her AI -powered talent agency. 'Sign my business's series a funding round with me,' read the caption of one TikTok, where she signed the £5 million deal with a pink fluffy pen. Just a few months later, she returned with a white fluffy pen to sign the €1.9 million round for Retrograde. Instead of relying on blog posts or LinkedIn updates, startups navigating a saturated market may find that a viral video is more likely to attract new customers—or even the right investor sliding into their DMs.
Yahoo
18-06-2025
- Business
- Yahoo
AI Innovators Worldwide Choose Oracle for AI Training and Inferencing
Fireworks AI, Hedra, Numenta, and Soniox experience accelerated performance and cost efficiency with Oracle Cloud Infrastructure AUSTIN, Texas, June 18, 2025 /PRNewswire/ -- AI innovators across the world are using Oracle Cloud Infrastructure (OCI) AI infrastructure and OCI Supercluster to train AI models and deploy AI inference and applications. Fireworks AI, Hedra, Numenta, Soniox, and hundreds of other leading AI innovators have selected OCI for its scalability, performance, cost efficiency, choice of compute instances, and control over where to run their AI workloads. As industries rapidly adopt AI to help drive innovation and efficiency, the AI companies that are providing these services require reliable, secure, and highly available cloud and AI infrastructure that enables them to quickly and economically scale out GPU instances. With OCI AI infrastructure, AI companies gain access to high-performance GPU clusters and the scalable computing power needed for AI training, AI inference, digital twins, and massively parallel HPC applications. "Among AI innovators, OCI has rapidly become the destination of choice for training and inferencing needs of all sizes," said Chris Gandolfo, executive vice president, Oracle Cloud Infrastructure and AI. "OCI AI infrastructure delivers ultra high-speed networking, optimized storage, and cutting-edge GPUs that AI companies rely on to power the next wave of innovation." Global AI Innovators Choose Oracle Fireworks AI is an inference platform that empowers developers and businesses to build highly optimized and production-ready generative AI applications, serving over 100 state-of-the-art open models in text, image, audio, embedding, and multi-modal formats. Fireworks AI uses OCI Compute bare metal instances accelerated by NVIDIA Hopper GPUs and OCI Compute with AMD MI300X GPUs to help it serve over two trillion inference tokens daily on its platform and scale its services globally. "Developers rely on Fireworks AI to integrate generative AI into their products, optimized for latency, throughput and cost per token," said Lin Qiao, co-founder and CEO, Fireworks AI. "With OCI AI infrastructure, we can deliver the ultra-fast response times and production-grade stability that developers expect. We're able to process AI workloads efficiently, minimize downtime, and help ensure AI applications run smoothly at scale so that our customers can focus on innovation without worrying about the underlying infrastructure." Hedra, an AI-driven video creation company, enables users to create videos with life-like characters. By deploying its multimodal foundation models for generative image, video, and audio on OCI Compute bare metal instances accelerated by NVIDIA Hopper GPUs, Hedra reduced its GPU costs, experienced faster training speeds, and reduced its model iteration time. "Creating expressive character videos at scale requires immense computational power and efficient multimodal processing," said Michael Lingelbach, founder and CEO, Hedra. "OCI handles our model training and inferencing across video, audio, and image data, while providing the rapid processing required for real-time character rendering and meeting the high storage demands of large datasets. This enabled us to release our latest model, Character-3, and content creation platform, Hedra Studio, quickly and without a hitch." Numenta is an AI technology company focused on maximizing the performance and efficiency of deep learning systems. By using OCI Compute bare metal instances accelerated by NVIDIA GPUs, Numenta gained access to a range of reliable and high-performance training instances, achieving faster training speeds and increased cycles of learning. "OCI provides the high-performance infrastructure and strong operational support we need to push the boundaries of AI without compromising speed or accuracy," said Dan Steere, CEO, Numenta. "With OCI, we've been able to confidently accelerate the development of our next-generation technology, taking significant steps forward in Efficient Intelligence™." Soniox, an AI company at the forefront of audio and speech AI, pioneers foundational AI models for audio, speech, and language comprehension. With its new universal multilingual speech AI model hosted on OCI, Soniox uses OCI Compute bare metal instances accelerated by NVIDIA Hopper GPUs to train its model to seamlessly recognize and understand speech across 60 languages in real-time with low latency and higher accuracy. "A high-performance infrastructure that provides improved accuracy, speed, and cost efficiency was top-of-mind when selecting a cloud provider to support our growth," said Klemen Simonic, founder and CEO, Soniox. "OCI gives us access to the latest AI innovations that allow us to push the boundaries of speech recognition and audio understanding while significantly reducing deployment time and operational costs." Additional Resources Learn more about Oracle Cloud Infrastructure Learn more about OCI AI infrastructure Learn more about OCI Generative AI Learn more about Oracle's AI strategy About OracleOracle offers integrated suites of applications plus secure, autonomous infrastructure in the Oracle Cloud. For more information about Oracle (NYSE: ORCL), please visit us at TrademarksOracle, Java, MySQL and NetSuite are registered trademarks of Oracle Corporation. NetSuite was the first cloud company—ushering in the new era of cloud computing. View original content to download multimedia: SOURCE Oracle
Business Insider
15-05-2025
- Entertainment
- Business Insider
Read the pitch deck an AI video startup behind viral baby podcast memes used to raise $32 million from A16z and others
A video of a baby interviewing a dog on a podcast went viral last month. No, it wasn't real. It was an AI-generated video created by comedian Jon Lajoie, who used Hedra, an AI video generation platform, to make the animation. Hedra's platform allows users to generate images, video, and audio with its web-based content creation studio. "Our model and technology focuses on the most controllable, compelling characters, whether that's a hyperrealistic human or an animated character or even an animal," Hedra's CEO, Michael Lingelbach, told Business Insider. On Thursday, Hedra announced that it raised a $32 million Series A fundraising round led by Andreessen Horowitz's Infrastructure fund. The round included returning investors such as A16z Speedrun, Abstract, and Index Ventures. Since its launch in 2024, the AI video startup has rapidly raised capital. In August, it announced a $10 million seed investment round. In March, Amazon 's Alexa Fund announced that it invested in the startup and several other AI companies. Hedra said it has raised a total of $44 million but has not disclosed a valuation. Competition in the generative AI is hot, with buzzy companies like Captions, HeyGen, Synthesia, and Runway building tech around video and avatars (Hedra specified that it is not an avatar company). "We're not trying to compete with Google Veo, we're not trying to compete with Sora," Lingelbach said. "We're focusing really firmly on building the best character models, and that's something that with this additional capital we can make another step function in doing." Hedra's Character-3 "omnimodal" model combines images, text, and audio to generate video. Creating a character with Hedra begins by uploading an image and then uploading audio that they've either already recorded (like a podcast) or generated using text-to-speech models like ElevenLabs. "Both voice and video are seeing rapid evolution right now," Lingelbach said. "We took a big leap forward on naturalness of expression with our current model." Hedra's platform is also users to integrate outside models like ElevenLabs, Google Veo, and Flux "all in one workflow," Lingelbach said. Hedra's core user base has been professional creators and marketers, Lingelbach said. "We're already seeing a massive influx of AI-generated content," Lingelbach said. "My Instagram and TikTok feed are filled with various memes and also more serious content now that's AI-generated." From comedy skits to faceless creator content to … talking babies, Hedra's already seen a wide range of use cases. Podcast content, particularly, has been a popular application of Hedra's tech. "It's not really something that we anticipated initially, but it definitely has been driving a lot of our usage," he said. In addition to the viral trend of AI baby-hosted podcasts that people have been creating using Hedra, others have used Hedra to create Studio Ghibli-style videos of the classic podcast interview clip. With its recent raise, Hedra plans to expand into more enterprise marketing applications, expand its team, and open an office in New York City. Note: Some slides have been redacted in order to share the deck publicly. Hedra Hedra is focused on storytelling and characters. The deck explains Hedra's 'omnimodal foundation model' that lets people quickly generate digital characters. Here's what the slide says: At Hedra, we've built the world's best character performance model that uniquely combines video, voice, motion, and emotion in a way never before possible. Hedra's Character-3 model is the world's first omnimodal foundation model in production. The only model that supports human, animated, and animal characters. And it works with any angle or framing. Built to prioritize efficiently scaling unified models The entire model was developed with a budget of under $2 million Hedra's customers range from everyday consumers to creators and marketers. The deck highlights Hedra's research team and its proprietary tech. Hedra Then the deck introduces the team. Here's what the slide says: We've assembled the best team to own this category — marrying deep research with AI-Native product design. Key Leadership Team: Michael Lingelbach: Founder / CEO Stanford PhD student of Fei-Fei Li and Jiajun Wu. Senior author of 3 real-time diffusion papers. Recipient of prestigious Stanford Graduate Fellowship. Hongwei Yi: Head of Research Former PhD Student of Michael Black, principal researcher behind first audio to video diffusion model to hit the market in the US. Wei Li: Research Lead Core contributor to Google Bard/Gemini, PaLM-2 and T5, with 8+ years experience at Google Brain/Deepmind. Jason Wilson: Head of Engineering Previously led engineering at Nava Benefits (Thrive-backed Series B startup) and engineering manager at Descartes Labs. Alan Guo: Chief of Staff MBA from Harvard Business School. Previously worked in growth & strategy at Disney, Jubilee Media, and Firework. Ramin Keene: Principal Engineer Hedra concludes its deck by saying 'we're just getting started.'
Yahoo
15-05-2025
- Business
- Yahoo
Hedra, the app used to make talking baby podcasts, raises $32M from a16z
People are using AI video generation tools to contribute to an unexpected new viral trend: podcasts featuring AI-generated talking babies. And one of the companies helping artists do this is Hedra. The startup, launched in 2023, offers a web-based video generation and editing suite powered by its Character-3 model, which lets users make videos with an AI-generated character as the focus, as well as transfer styles across images and audio. This is what people are using to make podcast videos like this one, in which an AI-generated dog talks about what it's like to live with a new baby in the house. We're not sure how much Hedra has benefited from this trend, but it's receiving ample investor attention nevertheless: the company on Thursday said it has raised $32 million in a Series A funding round led by Andreessen Horowitz's Infrastructure fund. Its previous investors are participating in the round, and a16z's Matt Bornstein will join the startup's board. Michael Lingelbach, the company's founder and CEO (pictured below), told TechCrunch the startup was inspired by the gap he noticed between companies like Synthesia, which let users superimpose AI-generated avatars over presentations, and startups like Runway, which provide video generation tools for creating short clips. "I thought what if we did something at the intersection of video generation and 3D characters, with long dialogues and better controllability," he said. Hedra launched its first video model in June 2024, and quickly attracted investor interest, securing $10 million in seed funding from Index Ventures, Abstract Ventures, and a16z speedrun. Earlier this year, Amazon also backed the company through its venture capital arm, Alexa Fund. Lingelbach noted that the launch of the Character-3 model in March was a big inflection point (shortly after the company signed its term sheet with a16z), and is now driving a lot of user growth. The startup wants to use fresh cash to train its next model, which it says enables better customization, as well as develop technology to let its AI-generated characters interact with users. The company is now focusing on attracting creators and prosumers, and said it has received inbound interest from marketing departments of enterprises as well. While Hedra's own model is centered around character movement and expression, the app lets you employ other models like Veo 2 and Kling for video generation; Flux, Imagen3, Sana, and Ideogram V2 for image generation; and audio models from ElevenLabs and Cartesia for voice generation or cloning. Hedra's competitors include Captions (also backed by a16z), which is focused more on smartphones; Greycroft-backed Cheehoo, which works with Hollywood studios to create animated features; Synthesia, and HeyGen. Hedra claims the videos generated with its platform have more expressive characters than those made using its competition. a16z's Bornstein thinks that as the AI-powered video generation space evolves, we will see more tools focusing on characters, motion, voice, editing and the like. "AI companies can produce amazing clips of environments and simple actions. But they can't generate meaningful dialogue or animation. It's not just about making a video, it's about making a story that resonates. This is largely down to the people and characters in the story. That's exactly what Hedra is building," he told TechCrunch in an emailed statement. This article originally appeared on TechCrunch at Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
