Haq's Musings

Riaz Haq writes this data-driven blog to provide information, express his opinions and make comments on many topics. Subjects include personal activities, education, South Asia, South Asian community, regional and international affairs and US politics to financial markets. For investors interested in South Asia, Riaz has another blog called South Asia Investor at http://www.southasiainvestor.com and a YouTube video channel https://www.youtube.com/channel/UCkrIDyFbC9N9evXYb9cA_gQ

Showing posts with label Nvidia. Show all posts

Wednesday, December 10, 2025

Pakistan Gets its First AI Data Center

Data Vault and Telenor Pakistan have launched the nation's first dedicated AI data center in Karachi. It is designed to support startups, researchers, and government agencies with high-performance computing and GPU-as-a-service offerings. It is equipped with more than 3,000 Nvidia's highest performance H100 and H200 GPUs for which the Trump Administration issued export licenses. These GPUs cost from $40,000 to $60,000 each, making the Nvidia chips the biggest chunk of the investment made in this AI data center. Other data centers in Pakistan also support AI workloads but this new data center in Karachi is specially designed for AI. It puts the country on a short list of only a handful of nations with locally hosted AI data centers. Pakistanis rank among the world's top five users of Artificial Intelligence (AI) tools, securing the fourth spot among 21 nations surveyed by the Schwartz Reisman Institute.

The local hosting of data in Pakistan ensures data sovereignty to comply with national data protection and security standards. It also achieves faster response times for queries. The data center runs entirely on solar power, making it a green data center solution. Additionally, the government of Pakistan has allocated 2,000 MW of power from the national grid for AI data centers.

One of the objectives of locally hosted AI data centers is to support Urdu language models (LLMs) trained to help Pakistani consumers who wish to use AI chatbots in local languages. A number of Urdu LLMs have already been developed in the country, including Alif and UrduLlama, both based on the open-source Llama-3.1-8B-Instruct architecture. Another model named UrduGPT is described as Pakistan's own large language model, UrduGPT is fine-tuned specifically for Urdu and regional languages of Pakistan using local datasets and cultural semantics to ensure relevance for native speakers.

Currently, Pakistan has 27 data centers located in Karachi, Lahore, and Islamabad, operated by PTCL, Multinet, and Cybernet. More are being built. Zong, a local mobile phone service operator owned by China Mobile, is building AI-driven cloud infrastructure in Pakistan. Indus Cloud and Huawei have a strategic partnership that aims to launch a next-generation cloud data center, incorporating energy-efficient Huawei technology. XDS and Al Nahal IT Park are partnering to build a liquid-cooled data center at the Al Nahal IT Park in Sindh province. Mari Petroleum Company Limited (MPCL), the state-owned oil and gas firm, is diversifying by forming a subsidiary, Mari Technologies, to build Tier III and Tier IV data centers in Islamabad and Karachi, with the 5MW Islamabad facility set for completion by early 2026.

Sunday, January 26, 2025

Chinese Lab's AI LLM Performance Shocks Silicon Valley

A Chinese Lab has sparked panic in Silicon Valley with the release of its first AI model that can outperform America's best despite being built more cheaply and with less-powerful chips, according to the US media reports. The lab called DeepSeek has recently unveiled a free, open-source large-language model (LLM) that it says took only two months and $5.5 million million to build, using reduced-capability chips from Nvidia called H800s. By comparison, the US-based OpenAI's closed LLM model cost $100 million to develop and train using the most advanced H100 chips from Nvidia. Open-source and free DeepSeek models can significantly help developing nations like Pakistan by providing affordable access to the latest AI technology, allowing them to develop solutions tailored to their specific needs without high costs.

DeepSeek, a small startup lab in China, has accomplished this feat despite the US technology export controls to slow down China's AI efforts. Former Google CEO Eric Schmidt is now acknowledging that China has narrowed or closed the AI technology gap with the United States.

In 2022 America banned the export of advanced chips to China, according to Economist magazine. Nvidia, a leading chipmaker, has had to design special downgrades to its products for the Chinese market. America has also sought to prevent China from developing the capacity to manufacture top-of-the-line chips at home, by banning exports of the necessary equipment and threatening penalties for non-American firms that might help, too.

The slower H800 chip was created by Nvidia to comply with export regulations that prevent the chipmaker from selling its high-end GPUs to China. Apparently, the limits imposed by Washington on Chinese engineers' access to the most advanced Nvidia chips forced them to develop a much more efficient model to achieve the same performance as their US counterparts. Other Chinese tech companies ranging from Alibaba and Huawei to TenCents are also working on their own multiple AI models, including LLMs.

DeepSeek has emerged from High-Flyer, a Chinese hedge fund started by 40-year-old Liang Wengfeng in 2015 to use AI to gain an edge in stocks-trading. Conducting fundamental research helped High-Flyer become one of the biggest quant funds in the country, according to The Economist magazine.