Microsoft will invest $3 billion to expand AI and cloud-computing infrastructure in India, CEO Satya Nadella announced during a conference in Bengaluru. The investment, the company’s largest expansion in the country, aims to strengthen its Azure cloud services and AI capabilities. Nadella also revealed plans to train 10 million people in AI by 2030, building on an earlier commitment to provide AI skilling opportunities for two million individuals by 2025, with a focus on smaller cities and rural areas.
India’s growing importance as a tech hub has attracted interest from major US technology firms, with recent visits from Nvidia’s Jensen Huang and Meta’s chief AI scientist Yann LeCun. Nadella met Prime Minister Narendra Modi to discuss technology, innovation, and Microsoft‘s ambitious plans for expansion in the country. India’s vast population and affordable internet access make it a key market for AI-driven growth.
Microsoft is making significant global investments in AI and cloud infrastructure, committing around $80 billion in fiscal 2025. More than half of that will be directed towards US data centers to support AI model training and cloud-based applications. With India positioned as a strategic market, Microsoft’s latest investment underscores the country’s growing role in the global AI ecosystem.
The Chinese technology powerhouse, Alibaba, has announced substantial price cuts of up to 85% for its large language models (LLMs), including the visual language model Qwen-VL. Designed to process and interpret both text and images, Qwen-VL is tailored for enterprise use, marking a departure from consumer-facing AI tools like ChatGPT. These discounts signal a competitive push to expand AI accessibility in the enterprise sector.
The move comes amid a broader race among Chinese tech giants to dominate the AI landscape. Companies like Tencent, Baidu, Huawei, and ByteDance have launched their own LLMs, aiming to capitalise on the growing demand for advanced AI solutions. Alibaba’s decision to focus on enterprise customers has already shown results, with its Qwen models adopted by over 90,000 businesses since May.
Analysts predict these price cuts could reshape global AI accessibility, enabling smaller firms and startups to leverage cutting-edge technology. Lower costs may allow traditional industries to modernise operations, while venture capital flows into supporting technologies are expected to further fuel innovation.
The global AI race is poised to accelerate into 2025, with Chinese companies playing a central role in advancing machine reasoning and practical applications. The intensifying competition could define the future of AI development, offering more use cases across diverse industries worldwide.
Taiwan Semiconductor Manufacturing Co (TSMC) has commenced mass production at its first factory in Kumamoto Prefecture, Japan. The facility manufactures 12 to 28-nanometer chips used in cars and image sensors, serving clients such as Sony Group and Denso Corp. Strengthening supply chains for critical goods is a priority for Japan, which views domestic chip production as vital for economic security amid geopolitical tensions.
TSMC plans a second factory in Kumamoto to produce advanced 6-nanometer chips, with construction set to begin by March 2025 and operations expected by late 2027. The Japanese government has pledged over 1 trillion yen in subsidies to support these initiatives, highlighting the strategic importance of reducing dependence on Taiwan’s chip supply.
Kumamoto Governor Takashi Kimura has also urged TSMC to consider a third plant in the prefecture, reflecting the region’s commitment to becoming a hub for semiconductor production. These developments underscore Japan’s determination to secure its technological future.
Prominent figures in technology are heavily investing in nuclear energy, viewing it as crucial for future innovation. OpenAI’s Sam Altman and Microsoft co-founder Bill Gates are spearheading initiatives in advanced nuclear technology, with Altman chairing Oklo, a company developing sustainable nuclear reactors.
Data centres, essential for AI and cloud technologies, have seen electricity demands surge by 50% since 2020, now accounting for 4% of US energy use. Projections indicate this figure could rise to 9% by 2030, emphasising the need for scalable, carbon-free energy solutions. Nuclear power offers a consistent energy supply, unlike solar or wind, making it an attractive choice.
Microsoft has committed to reviving the Three Mile Island reactor by 2028, aiming to meet the energy needs of its growing AI operations. Experts, however, caution that tech-driven nuclear investments may prioritise corporate demands over broader public benefits.
Oklo and similar ventures highlight the increasing convergence of technology and energy, as industry leaders strive to support AI advancements sustainably. The debate continues on whether these moves truly serve societal needs or primarily benefit the tech sector.
The increasing number of data centres powering AI could pose significant challenges for the United States power grid, as reported by Bloomberg. Findings indicate a connection between data centre activity and ‘bad harmonics,’ a term describing electrical power distortions that can damage appliances, heighten fire risks, and lead to power outages.
Bloomberg’s analysis, using data from Whisker Labs and DC Byte, revealed that over half of homes with the worst power distortions are located within 20 miles of active data centres. AI-driven centres, with their unpredictable energy needs, exacerbate these grid strains, pushing infrastructure beyond its designed limits.
Experts, including Aman Joshi of Bloom Energy, warn that no current grid can handle such intense load fluctuations from multiple data centres. While some utility companies question these findings, the report underscores the urgent need to address the interplay between technological expansion and energy stability.
As concerns grow over the impact of smartphones on children, several European countries are implementing or debating restrictions on their use in schools. France, for example, has prohibited phones in primary and secondary schools since 2018 and recently extended the policy to include ‘digital breaks’ at some institutions. Similarly, the Netherlands and Hungary have adopted bans, with exceptions for educational purposes or special needs, while Italy, Greece, and Latvia have also imposed restrictions.
The debate is fueled by studies showing that smartphones can distract students, though some argue they can also be useful for learning. A 2023 UNESCO report recommended limiting phones in schools to support education, with more than 60 countries now following similar measures. However, enforcement remains a challenge, as some reports suggest that many students still find ways to use their devices despite the bans.
Experts remain divided on the issue. While some highlight the risks of distraction and mental health impacts, others emphasise the need for balance. ‘Banning phones can be beneficial, but we must ensure children have adequate alternatives for education and communication,’ said Ben Carter, a professor of medical statistics at King’s College London.
The trend reflects broader concerns about screen time among children, with countries like Sweden and Luxembourg calling for clearer rules to promote healthier digital habits. While opinions differ, the growing movement underscores a collective effort to create focused, engaging, and healthier learning environments.
Losing the ability to speak can feel overwhelming, but Apple’s innovative Personal Voice and Live Speech features offer a transformative solution. Designed for individuals at risk of losing their voice, these tools allow users to create a synthesised version of their own voice, preserving their unique communication style even when speaking becomes difficult.
Personal Voice works by recording specific phrases on an iPhone, iPad, or Mac. These recordings are encrypted and stored securely on the user’s device, ensuring privacy. Once created, the personalised voice can be used across Apple devices running iOS 17, iPadOS 17, or macOS Sonoma, enabling seamless communication.
Setting up Personal Voice is simple and requires a quiet space for accurate recording. Apple’s features empower individuals dealing with progressive conditions, recovery from injuries, or anyone seeking a communication backup, underscoring technology’s ability to enhance accessibility and maintain personal expression.
AI is transforming education for students with disabilities, offering tools that level the playing field. From reading assistance to speech and language tools, AI is enabling students to overcome learning barriers. For 14-year-old Makenzie Gilkison, who has dyslexia, AI-powered assistive technology has been life-changing, allowing her to excel academically and keep pace with her peers.
Schools are increasingly adopting AI for personalised learning, balancing its benefits with ethical considerations. Tools like chatbots and text-to-speech programs enhance accessibility while raising concerns about over-reliance and the potential for misuse. Experts emphasise that AI should support, not replace, learning.
Research and development are advancing rapidly, addressing challenges like children’s handwriting and speech impediments. Initiatives such as the National AI Institute for Exceptional Education aim to refine these tools, while educators work to ensure students and teachers are equipped to harness their potential effectively.
Chinese AI firm DeepSeek has unveiled DeepSeek V3, a groundbreaking open-source model designed for a range of text-based tasks. Released under a permissive licence, the model supports coding, translations, essay writing, and email drafting, offering developers the freedom to modify and deploy it commercially.
In internal benchmarks, DeepSeek V3 outperformed major competitors, including Meta’s Llama 3.1 and OpenAI’s GPT-4o, especially in coding contests and integration tests. The model boasts an impressive 671 billion parameters, significantly exceeding the size of many rivals, which often correlates with higher performance.
DeepSeek-V3!
60 tokens/second (3x faster than V2!) API compatibility intact Fully open-source models & papers 671B MoE parameters 37B activated parameters Trained on 14.8T high-quality tokens
DeepSeek V3 was trained on a dataset of 14.8 trillion tokens and built using a data centre powered by Nvidia H800 GPUs. Remarkably, the model was developed in just two months for a reported $5.5 million—far less than comparable systems. However, its size and resource demands make it less practical without high-end hardware.
Regulatory limitations influence the model’s responses, particularly on politically sensitive topics. DeepSeek, backed by High-Flyer Capital Management, continues to push for advancements in AI, striving to compete with leading global firms despite restrictions on access to cutting-edge GPUs.
AI startups specialising in sales development representatives (SDRs) are experiencing rapid growth as businesses embrace new technologies to streamline outreach. These startups, leveraging large language models (LLMs) and voice technology, automate tasks like crafting personalised emails and placing calls to potential customers. This sector has seen an unprecedented surge, with multiple companies achieving notable success in a short span, according to Shardul Shah of Index Ventures. However, investors remain cautious about whether this trend will yield lasting results or fade once the novelty wears off.
The appeal of AI SDRs is particularly strong among small and medium-sized businesses, which find it easier to experiment with these tools. Arjun Pillai, founder of Docket, attributes the popularity to declining reply rates for traditional cold emails, prompting businesses to explore AI-driven solutions. Startups like Regie.ai, AiSDR, and 11x.ai, as well as incumbents like ZoomInfo, are vying for market share, boasting impressive revenue growth. Yet, as Tomasz Tunguz of Theory Ventures noted, some businesses report that while AI SDRs generate substantial leads, they don’t necessarily translate into higher sales, highlighting a gap in effectively integrating AI into sales strategies.
Despite the enthusiasm, the rise of AI SDRs faces significant challenges. Industry leaders such as Salesforce and HubSpot, which control vast customer data, could introduce similar AI features, potentially outpacing smaller startups. Investors also point to cautionary tales like Jasper, a copywriting AI startup that stumbled after the launch of ChatGPT, emphasising the uncertainty surrounding the longevity of AI adoption in sales. For now, the potential of AI SDRs to revolutionise sales processes is undeniable, but their ability to sustain growth and deliver tangible results remains to be seen.
