Venture Hub

People posting on X/Twitter that I'm wrong. But they just totally misunderstood my arguments. The comments are even worse! I don't comment on X/Twitter anymore, so they'll never know ?? https://lnkd.in/e75XJhgH

Thailand’s ‘data centre rush’ is starting, as Malaysia strains to keep pace with surging demand: Macquarie Analysts at financial services company choose Bangkok-based Gulf Energy as their top pick Summary: The Southeast Asian data center market is experiencing unprecedented growth driven by global tech companies racing to meet the demand for infrastructure and computing power fueled by AI. While Malaysia continues to expand its data center capacity, it is struggling to keep pace with surging demand due to power constraints. Johor, a key data center hub, recently rejected 30% of applications over the past five months to conserve resources and prioritize local benefits. Meanwhile, Thailand is emerging as a strong competitor in the region's "data center rush." With ample power infrastructure and supportive regulations, Thailand is well-positioned to attract investments from global players like Amazon and Google, particularly in the Eastern Economic Corridor. The shift to a "site-to-power" model—building data centers near robust power sources—is further enhancing Thailand’s appeal as a destination for hyperscaler facilities. Macquarie analysts highlighted Gulf Energy as their top pick in Thailand, citing its dual role in developing data centers and supplying power to hyperscalers. In Malaysia, they identified YTL Power and Tenaga Nasional Berhad (TNB) as key players, with YTL Power's AI data center set to launch in mid-2025, featuring early access to Nvidia’s GB200 servers. As AI workloads grow exponentially, Southeast Asia, led by Thailand and Malaysia, is cementing its status as a critical data center hub, attracting multibillion-dollar investments while addressing power and sustainability challenges. #DataCenters #AI #SoutheastAsia #Thailand #Malaysia #DigitalInfrastructure #Hyperscalers #Sustainability #PowerInfrastructure #TechGrowth #CloudComputing

?? Exciting Start to a New Chapter in Thailand!?? We at VentureHub are thrilled to have initiated meaningful discussions with the Thai Defence Information and Security Department to explore collaborations in Artificial Intelligence. This could pave the way for advancing national security, enhancing defense capabilities, and strengthening cybersecurity, as AI continues to transform the landscape of security and defense worldwide. At VentureHub, we are on a mission to support AI sovereignty. From AI infrastructure to supporting local AI applications from all sectors, we are at the forefront of this critical dialogue and we look forward to uncovering opportunities that will benefit Thailand, the region, and the global community. Stay connected for updates as we push the boundaries of AI innovation!?? #AI #Innovation #NationalSecurity #Cybersecurity #Defense #VentureHub

A Special Message to Schools and Educational Institutions Every language model is deeply influenced by the data it is trained on. Therefore, it is essential for government offices and public institutions serving the public to develop language models aligned with the culture, values, and ethics of their communities. ? A Special Message to Schools and Educational Institutions: Please do not use these models with your students or audience without evaluating and ensuring their relevance to your community’s context and needs. Proper review guarantees a positive and respectful educational and interactive experience. #AI #CulturalSensitivity #Ethics

Amazing. You can now convert images of equations into LaTeX code with simple Python library. Use the command line tool by calling pix2tex: ?????? ?????????????? "??????????????[??????]" ???????? ?????? ???????????? ?????????? ???????? ??????????????.?????? ???????????? ???????????????? ?????? = ??????????.????????('????????/????/??????????.??????') ?????????? = ????????????????() ??????????(??????????(??????)) Repo: https://lnkd.in/d8eTH-iq ?? Repost this if you found it useful. ↓ Are you technical? Check out?https://AlphaSignal.ai?to get a daily summary of breakthrough models, repos and papers in AI. Read by 200,000+ devs.

Game changer for Gen AI evaluations: Galileo just introduced Luna, a family of Eval Models 97% cheaper, 11x faster, and 18% more accurate than GPT-3.5. Current RAG methods struggle with accurate, low-latency, low-cost hallucination detection. Luna, a fine-tuned DeBERTA-large encoder, solves this by detecting hallucinations, enhancing security, and providing customizable evaluations without ground truth data. Features: ? Cost effective: 97% cheaper than GPT-3.5 ? Fast: 11x faster than GPT-3. ? Accurate: 18% more accurate than GPT-3.5 ? No Ground Truth Data Needed: Simplifies deployment and maintenance ? Customizable: Quickly fine-tune for your specific evaluation requirements ? Robust Security: Detect hallucinations, prevent prompt attacks, and enforce data privacy ? Extended Context Support: Handles 16k+ context length seamlessly Paper: https://lnkd.in/g6xDqgWw Blog Post: https://lnkd.in/ggZpuVDD

Thank you Jay for sharing us QCT high performance product series, surely the most wanted GB200 machine. And thank you Marcus Yang, PE for making the introduction. We look forward to our collaboration in this AI space.

https://lnkd.in/e7NPhcX7

Maintaining such a costly support to AI infrastructure over the long term will be a key challenge for a government, especially if it wants domestic startups to compete internationally. We are on a mission to allow corporations and users from government sectors to unleash the power of AI through private-public sector collaboration

“Tokyo-based Sakana AI, founded by former Google Research members, is using evolution-inspired techniques to ‘breed’ multiple generations of LLMs.” Source: https://lnkd.in/gfy-_g46 https://lnkd.in/gyR8aDJq