Microsoft Report Highlights DeepSeek AI Growth and Scale AI Findings

A recent Microsoft report from January 8, 2026, highlights Chinese AI startup DeepSeek's increasing popularity in developing nations. Founded in 2023, DeepSeek offers free and open-source AI models, making advanced AI more accessible and affordable in price-sensitive regions like Southeast Asia, Africa, and Latin America. Its R1 model, released in January 2025, proves more cost-effective than similar offerings from OpenAI. This trend helps narrow the AI adoption gap, with generative AI use in developing countries increasing by 2.4 percentage points in late 2023, surpassing advanced economies. However, developed nations like the US and Germany limit DeepSeek's use due to security concerns, and its political answers differ from US models. Beyond global adoption, companies are seeing tangible benefits from AI. Personio's Chief Revenue Officer, Philip Lacor, transformed the company's sales strategy in just six months, deploying over 400 AI assistants. This initiative dramatically reduced sales research time from two hours to 15 minutes per representative, leading to 140 meetings booked in a single week. Meanwhile, the U.S. Department of Energy is investing over $320 million to advance AI capabilities, including the new Learning with AI and Data Science (LEADS) Institute. Virginia Tech professor Yulia R. Gel is a key member, focusing on developing algorithms for large data and real-time information using digital twins. Despite these advancements, current AI systems still face significant limitations. A study by Scale AI and the Center for AI Safety found that leading AI tools like ChatGPT completed only 2.5% of real work assignments, struggling with tasks requiring long-term memory or visual understanding, such as creating detailed floor plans. In the legal sector, Pennsylvania judges are encountering "AI hallucinations" in court documents, with at least 13 cases in 2025 containing confirmed AI-generated errors, including fake citations. Furthermore, CES 2026 showcased numerous "AI-powered" gadgets, like Glyde smart hair clippers and SleepQ's "AI-upgraded pharmacotherapy," raising questions about the genuine utility of AI in some consumer products. The investment landscape for AI is also evolving. HSBC remains optimistic about major tech firms like Nvidia, Alphabet (Google's parent company), and Microsoft, giving them "Buy" ratings, but warns that AI investments are becoming tougher. While AI boosted many sectors in 2025, the market mood shifted as large spending on AI infrastructure outpaced quick returns. For 2026, HSBC anticipates strong cloud demand but expects ongoing shortages in capacity, power, and chips. On the regulatory front, Vermont State Representative Monique Priestley urges state-level action on AI, emphasizing that states are crucial for creating new laws to protect citizens against discrimination and abuse, especially as technology outpaces federal action. NVIDIA also launched TensorRT Edge-LLM on January 8, 2026, a new tool to speed up AI models for cars and robots on edge devices, already being used by companies like Bosch and MediaTek.

Key Takeaways

• DeepSeek AI, a Chinese startup, is rapidly gaining popularity in developing nations due to its free, open-source, and affordable models, as reported by Microsoft.
• DeepSeek's R1 model, released in January 2025, is more cost-effective than similar offerings from OpenAI, contributing to its adoption in price-sensitive regions.
• Developed nations like the US and Germany limit DeepSeek's use due to security concerns, despite its growth in China, Russia, and Iran.
• NVIDIA launched TensorRT Edge-LLM on January 8, 2026, to accelerate AI models for cars and robots on edge devices, with companies like Bosch and MediaTek already using it.
• A study by Scale AI and the Center for AI Safety found that leading AI systems, including tools like ChatGPT, completed only 2.5% of real work assignments, indicating they are far from replacing human jobs.
• The U.S. Department of Energy is investing over $320 million in AI capabilities, establishing the LEADS Institute to develop advanced algorithms for data exploration and real-time information.
• Personio's CRO, Philip Lacor, successfully deployed over 400 AI assistants, cutting sales research time from two hours to 15 minutes and booking 140 meetings in one week.
• Pennsylvania judges have encountered "AI hallucinations" or errors in at least 13 court cases in 2025, highlighting concerns about AI accuracy in legal documents.
• HSBC maintains "Buy" ratings for major tech firms like Nvidia, Alphabet (Google), and Microsoft, but warns that AI investments are becoming more challenging due to infrastructure spending outpacing quick returns.
• Vermont State Representative Monique Priestley advocates for state-level AI regulation, arguing that states are essential for creating timely laws to protect citizens from AI's impact on services like housing, jobs, and healthcare.

DeepSeek AI grows in developing nations Microsoft reports

A Microsoft report from January 8, 2026, shows that Chinese AI startup DeepSeek is becoming popular in many developing countries. This trend could help close the gap in AI use between advanced and developing economies. DeepSeek, founded in 2023, offers free and open-source AI models, making it affordable for users in price-sensitive regions. Its advanced R1 model, released in January 2025, is more cost-effective than OpenAI's similar model. However, developed nations like the US and Germany limit DeepSeek due to security concerns, and its political answers differ from US models.

DeepSeek AI thrives in developing countries Microsoft finds

A Microsoft report titled "The State of AI in 2023" reveals that Chinese AI startup DeepSeek is gaining popularity in many developing nations. This growth helps narrow the gap in AI adoption between advanced and developing economies. Generative AI use in developing countries increased by 2.4 percentage points in late 2023, outpacing advanced economies. DeepSeek's success comes from its open-source, affordable models and its efforts to localize products for regions like Southeast Asia, Africa, and Latin America. This approach makes advanced AI more accessible and helps bridge the global digital divide.

DeepSeek AI grows in developing nations Microsoft reports

A Microsoft report from January 8, 2026, indicates that Chinese AI startup DeepSeek is gaining popularity in many developing countries. This trend could help close the gap in AI use between advanced and developing economies. DeepSeek, founded in 2023, offers free and open-source AI models, making it affordable for users in price-sensitive regions. Its advanced R1 model, released in January 2025, is more cost-effective than OpenAI's similar model. While DeepSeek's adoption surged in China, Russia, and Iran, developed nations like the US and Germany limit its use due to security concerns.

Security experts advise buying AI solutions for clear results

A security expert suggests that businesses should stop buying AI for its buzzwords and instead focus on the real outcomes it provides. AI will soon become a standard part of technology, working silently in the background. Companies should invest in AI that helps their teams with tasks like managing alerts, investigating issues, and creating reports. The article highlights three key areas where AI offers true value: summarizing complex information, navigating large amounts of data with natural language, and prioritizing tasks. This approach ensures AI enhances human decision-making without taking away control, especially in critical industries.

Personio CRO built AI sales strategy in six months

Philip Lacor, the Chief Revenue Officer at Personio, successfully transformed the company's sales strategy using AI in just six months. Starting in May 2024, Personio moved from basic AI use to deploying over 400 AI assistants. This change dramatically cut sales research time from two hours to 15 minutes per representative and helped book 140 meetings in one week. Lacor emphasized the need for both top-down leadership and bottom-up experimentation for AI initiatives to succeed. He also highlighted the importance of a cross-functional team, including data, operations, and business units, to ensure AI tools meet real business needs.

Virginia Tech professor joins national AI science project

The U.S. Department of Energy is investing over $320 million to advance AI capabilities, including a new initiative called the Learning with AI and Data Science LEADS Institute. Yulia R. Gel, a statistics professor at Virginia Tech, is a key member of this new project. The LEADS Institute will develop new algorithms to explore large data and extract real-time information using digital twins. Gel's research focuses on graph learning and AI. Panos Stinis will direct the institute, which aims to connect AI experts with scientists to create advanced algorithms.

New study shows AI cannot yet replace human jobs

A new study from Scale AI and the Center for AI Safety reveals that current AI systems are far from replacing human jobs. Researchers tested leading AI tools like ChatGPT on hundreds of real work assignments. The best AI system successfully completed only 2.5% of these projects. AI struggled with tasks like creating detailed floor plans or producing high-quality 3D product videos. Major limitations include AI's lack of long-term memory and its difficulty with visual understanding. The study suggests that predictions about AI automating large parts of the workforce are still far from reality.

Vermont lawmaker urges state action on AI regulation

Vermont State Representative Monique Priestley argues that Vermont must act quickly to regulate artificial intelligence, despite a White House executive order targeting state-level AI laws. Priestley believes states are crucial for creating new laws, especially since technology advances faster than federal action. AI systems already impact essential services like housing, jobs, and healthcare, making state-level protections against discrimination and abuse vital. She criticizes the federal government for threatening states with grant losses if they pursue their own AI regulations. Priestley emphasizes that states across the country are working together to protect citizens and ensure AI benefits the public.

HSBC warns AI investments are becoming tougher

HSBC remains optimistic about Big Tech companies due to artificial intelligence, but warns that investing in AI is becoming more challenging. The bank gives "Buy" ratings to major tech firms like Nvidia, Alphabet, Microsoft, Amazon, Meta Platforms, Oracle, and Broadcom. However, HSBC is more cautious about Apple and CoreWeave. While AI boosted many sectors in 2025, the market mood shifted as large spending on AI infrastructure outpaced quick returns. For 2026, HSBC expects strong cloud demand but anticipates ongoing shortages in capacity, power, and chips.

NVIDIA launches new AI tool for cars and robots

NVIDIA introduced TensorRT Edge-LLM on January 8, 2026, a new tool designed to speed up AI models for cars and robots. This lightweight software helps Large Language Models and Vision-Language Models work efficiently on edge devices. It offers features like in-place weight updates and dynamic input shapes, ensuring low and predictable latency for critical applications. Companies like Bosch, ThunderSoft, and MediaTek are already using TensorRT Edge-LLM in their products, showcasing it at CES 2026. The tool is available for NVIDIA Jetson AGX Thor DevKit and is part of the NVIDIA DriveOS package.

CES 2026 showcases questionable AI gadgets

At CES 2026, many gadgets claimed to use AI in questionable ways, according to Dominic Preston of The Verge. Examples include Glyde smart hair clippers with an AI coach that offers hairstyle advice. SleepQ introduced "AI-upgraded pharmacotherapy" which uses smartwatch data to suggest when to take sleeping pills. The Fraction stick vacuum uses "Neural Predictive AI" to monitor its own performance and suggest replacement parts. Another device, the Fraimic AI art frame, generates images using OpenAI's GPT Image 1.5 based on voice commands, costing $399 for 100 images annually.

Pennsylvania judges find AI errors in court documents

Pennsylvania judges are discovering "AI hallucinations" or errors in court documents, raising concerns about legal accuracy. In 2025, at least 13 cases in Pennsylvania contained confirmed or implied AI-generated mistakes, mostly from individuals representing themselves. University of Pittsburgh law professor David A. Harris warns that fake citations distort the law. Commonwealth Court Judge Matthew Wolf recently confronted attorneys Thomas Breth and Thomas W. King III about an error-filled brief, which included misquotes and non-existent citations. The attorneys acknowledged their responsibility, calling the situation "incredibly embarrassing."

Sources

• DeepSeek's AI gains traction in developing nations, Microsoft report says
• DeepSeek's AI gains traction in developing nations, Microsoft report says
• DeepSeek's AI gains traction in developing nations, Microsoft report says
• Security Think Tank: Stop buying AI, start buying outcomes
• How Personio’s CRO Built an AI-Powered Go-To-Market in Just 6 Months: 5 Lessons and 5 Mistakes
• Statistician part of new effort to advance scientific machine learning
• Analysis | Can AI do your job? See the results from hundreds of tests.
• Column: Artificial Intelligence doesn’t wait; Vermont can’t either
• HSBC Says The AI Trade Is Getting Harder
• Accelerating LLM and VLM Inference for Automotive and Robotics with NVIDIA TensorRT Edge-LLM
• Most dubious uses of AI at CES 2026
• Judges are identifying suspected AI hallucinations in Pa. court cases