Ad
Ad
Ad
Short

Saudi Arabia and the United Arab Emirates (UAE) are competing to become the leading AI nation in the Middle East, investing heavily in data centers, according to Bloomberg. The UAE currently has more data center capacity at 235 megawatts compared to Saudi Arabia's 123 megawatts, but Saudi Arabia is planning a larger expansion. By comparison, Germany is expected to have 1,060 megawatts of capacity. A PwC study predicts that AI could contribute $135 billion to Saudi Arabia's GDP and $96 billion to the UAE's by 2030. While the UAE is seen as a technology leader, Saudi Arabia's policies are deterring investors. However, Saudi Arabia offers cloud providers more favorable energy prices. The biggest challenge is operating water-intensive data centers in the desert. U.S. data center operator Equinix is building its fourth data center in Dubai and has plans for further regional expansion. OpenAI CEO Sam Altman is reportedly in talks with the UAE about an AI chip factory.

Short

Microsoft proposed to the U.S. Department of Defense in 2023 that it use OpenAI's DALL-E AI image generator to develop software for military operations, according to internal company documents reviewed by The Intercept. A Microsoft presentation showed how the Pentagon could potentially use OpenAI tools like ChatGPT and DALL-E for tasks like document analysis and machine maintenance. One Microsoft document suggested using images generated by DALL-E to train battlefield management systems, possibly as synthetic training data. In a statement to The Intercept, Microsoft said it had made this suggestion but had not actually implemented it. OpenAI was not involved in creating Microsoft's presentation and has not sold any of its tools directly to the U.S. military. While Microsoft is actively pursuing military projects, OpenAI has only recently opened up to that possibility, with the limitation that its technologies cannot be used to develop weapons or harm humans.

Ad
Ad
Short

Paris-based AI startup Mistral has released Mixtral-8x22B MoE, a new open language model, via a torrent link. An official announcement with more details will follow later. According to early users, the model offers 64,000 token context windows and requires 258 gigabytes of VRAM. Like the Mixtral-8x7B, the new model is a mixture-of-experts model. You can try out the new model at Together AI.

Image: Together AI | X.com
Ad
Ad
Google News