The AI Book
    Facebook Twitter Instagram
    The AI BookThe AI Book
    • Home
    • Categories
      • AI Media Processing
      • AI Language processing (NLP)
      • AI Marketing
      • AI Business Applications
    • Guides
    • Contact
    Subscribe
    Facebook Twitter Instagram
    The AI Book
    Daily AI News

    Tokyo startup Sakana AI lands $30M to forge new path with compact AI models

    16 January 2024No Comments3 Mins Read

    [ad_1]

    Sakana AI, a Tokyo-based artificial intelligence startup co-founded by two notable ex-Google engineers, announced today that it has raised $30 million in seed funding from high-profile technology investors. The young company, founded just last year, aims to take a different approach to AI by developing smaller, more efficient models inspired by nature. 

    The seed round was led by Lux Capital, which has invested in pioneering AI companies like Hugging Face, and also Khosla Ventures, known for investing in OpenAI way back in 2019. Japanese tech giants Sony, NTT and KDDI also participated in the round, marking a vote of confidence in Sakana from major domestic players.

    Sakana, which means “fish” in Japanese, takes its inspiration from the collective behaviors of animal groups like schools of fish and flocks of birds. The startup believes that getting smaller AI models to work together efficiently can be more effective than training single gigantic models on massive datasets, an approach that many leading AI labs have pursued.

    Sakana co-founders David Ha and Llion Jones previously led AI research groups at Google. Jones co-authored a landmark paper on the Transformer model in 2017 that underpins chatbots like OpenAI’s ChatGPT today.  

    Sakana’s approach stands out at a time when the AI field has fixated on scaling up models to beat benchmarks through sheer size. This tactic has produced impressive results, but also led to backlash over the computing resources and environmental impact required to train and run colossal AI systems.

    Sakana execs argue that large models become inefficient as they balloon in size, while smaller, specialized models can collaborate to match the capabilities. The startup likens it to how groups of people each with distinct skills outperform lone polymaths at complex jobs.

    The founding team’s pedigree and alternative vision were enough to attract top Silicon Valley and domestic Japanese investors to fund Sakana’s Tokyo lab after just one year of operating independently. With fresh capital and partnerships with the likes of NTT, the company will look to staff up and further develop its nature-inspired AI techniques.

    Early-stage backing from U.S. and Japanese tech heavyweights signals confidence that Sakana could pioneer a new AI paradigm from Asia, while giving Japan influence in a strategic technology where the U.S. and China have dominated so far.

    VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.



    [ad_2]

    Source link

    Previous ArticleSynopsys agrees to buy system design firm Ansys for $35B
    Next Article WEF 2024 Report: Cybersecurity at the forefront, zero trust seen as critical for trust rebuilding
    The AI Book

    Related Posts

    Daily AI News

    Adobe Previews New GenAI Tools for Video Workflows

    16 April 2024
    Daily AI News

    Exciting Updates From Stanford HAI’s Seventh Annual AI Index Report

    15 April 2024
    Daily AI News

    8 Reasons to Make the Switch

    15 April 2024
    Add A Comment

    Leave A Reply Cancel Reply

    • Privacy Policy
    • Terms and Conditions
    • About Us
    • Contact Form
    © 2025 The AI Book.

    Type above and press Enter to search. Press Esc to cancel.