Arda Translate is a private voice translation application designed to run entirely on iOS devices. Its core feature is performing both speech recognition and translation on-device, eliminating the need to send conversation data to cloud services. This design ensures user privacy, making it suitable for sensitive conversations where data security is paramount.

The app supports a wide range of languages, including English, Chinese, Spanish, French, German, Portuguese, Italian, Japanese, and Korean. Users need to download a large translation model during the initial setup, for which a Wi-Fi connection is recommended. Arda Translate operates without requiring an account, displaying ads, or demanding a subscription, offering a one-time purchase model.

A notable feature is its live two-sided conversation view, which allows two individuals to see the conversation translated into their respective languages on the same screen. This functionality is ideal for everyday interactions in settings like travel, family discussions, or small businesses, though it is not recommended as the sole translation source for critical situations such as medical or legal contexts. Developer CHRISTOPHER JOHN BORDELEAU explicitly states that the app collects no user data, reinforcing its strong privacy stance.

👉 Read the full article on App Store

China has made significant strides in catching up with the American AI industry, largely by focusing on open-source models. However, Beijing is now considering limiting foreign access to its top AI systems, engaging in discussions with major Chinese tech developers like Alibaba, ByteDance, and Z.ai. These potential restrictions would apply to both open-source and proprietary models, and could also impact funding for Chinese AI startups.

Chinese AI models have been closing the performance gap with advanced U.S. models from companies like OpenAI and Anthropic. Notably, Z.ai's GLM-5.2 reportedly shocked Silicon Valley by achieving comparable performance to proprietary U.S. models at a fraction of the training cost. This technical progress, coupled with the emergence of Anthropic's Mythos, a cybersecurity AI model, has intensified China's resolve to protect its domestic AI industry.

Beijing has pledged to invest approximately 2 trillion yuan ($294 billion) in domestic data center development over the next five years, leading many Chinese AI developers to shift from Nvidia chips to local silicon providers. This strategic pivot underscores the belief that AI is a critical geopolitical asset, with the nation that masters AI poised to dominate the 21st-century world order. Author and historian Yuval Noah Harari refers to this growing technological and ideological divide as the “Silicon Curtain,” suggesting it could hinder cooperation between the U.S. and China.

👉 Read the full article on Gizmodo

NVIDIA is challenging conventional data center cooling paradigms with a new approach. The company has unveiled its next-generation Rubin infrastructure, featuring 100% liquid-cooled “AI factories.” NVIDIA claims that in this system, the coolant can enter the chips at 45°C and exit at approximately 55°C without compromising processor performance.

Unlike traditional data centers that consume vast amounts of air and energy to cool entire rooms, NVIDIA's DSX design efficiently captures heat by directly supplying coolant to the chips. This liquid, composed of 75% water and 25% propylene glycol, circulates in a closed loop, eliminating the need for fans within the system. This design increases the data center's flexibility regarding ambient temperatures, allowing operation without reliance on cold air.

This liquid cooling system has the potential to reduce water consumption for cooling to near zero, saving millions of gallons annually compared to conventional cooling tower systems. Furthermore, with cooling accounting for up to 40% of a data center's electricity consumption, NVIDIA estimates significant energy cost reductions by raising target temperatures. As AI demand is projected to double data center electricity consumption by 2030, this efficient cooling technology becomes strategically vital.

👉 Read the full article on Gizmodo en Español