Shiva's Blog #4 on topics of AI, Cloud, Automation, Technology and Geopolitics

This is 4th the series of blogs I am writing on topics of AI, Cloud, Automation (I abbreviate it as AiClAu), Technology and Geopolitics. The 3rd blog was published in LinkedIn on Nov 16th, 2024. I review relevant articles, news items, white papers, product releases, events and summarise them in these blogs to the benefit of Tech and Non-tech audiences.  Titles and URL of these sources may be listed at the end of each blog.

On 20th Jan this year, a mainland China startup company DeepSeek released its Large Language Model (LLM) R1. They reportedly developed the model by spending less than US$6 million on computing power using Nvidia H800 chips. This release took the AI world by storm as American companies typically spent hundreds of millions / billions of $ for developing their LLMs.

DeepSeek App become one of the top free Apps in Apple's App store. Consequently, there was a massive tech stocks sell-off on Jan 27th, 2025 (With Nvidia declining by 17% and losing approximately $600 billion in market capitalization on one day).

DeepSeek is a Hangzhou-based startup founded in May 2023 by Liang Wenfeng, also co-founder of quantitative hedge fund High-Flyer. DeepSeek provides multiple services for its models, including a web interface, mobile application and API access. DeepSeek now presents a major challenge / opportunity to US tech giants like OpenAI, Microsoft, Meta, Google, Grok and others. One major difference is DeepSeek offering its model to the Open source world unlike OpenAI paid subscription.

Being based out of China, DeepSeek caused huge Geopolitical concerns to U.S. technological dominance in AI space. US and China now consider each of themselves as competitors in Economy, Culture, Technology, Military, and Soft power space. Tech investor Marc Andreessen called this DeepSeek release as AI's "Sputnik moment," comparing it to the Soviet Union's space race breakthrough in the 1950s.

Let us know examine the events from the perspective of Microsoft, the largest company in the world in terms of market capitalization (as of June 2025).

Under Microsoft CEO Satya Nadella’s leadership, Microsoft made a bold and early bet on OpenAI, the research lab behind breakthrough AI models like GPT-3 and ChatGPT. Beginning in 2019, Microsoft invested billions of dollars into OpenAI, providing not just capital but also essential cloud computing resources through its Azure platform. So, Microsoft uses OpenAI technology in its products as the underlying AII model but also provides its Azure services to OpenAI as a vendor. This multifaceted partnership enabled OpenAI to scale its research and development efforts, while giving Microsoft privileged access to some of the world’s most advanced AI technologies.

Microsoft (MS) hedges its AI bet smartly by also providing its own in-house models and other external models in addition to OpenAI model. Satya Nadella is a very suave business leader who built the Azure cloud business for Microsoft before becoming its third CEO on 4th Feb, 2014. MS has invested more than $13 billion in OpenAI so far and has committed close to US$ 80 billion on building Data centers powering AI.

He and his team were in the World Economic Forum (WEF) in Davos in Jan 2025 when the DeepSeek release happened. Instead of seeing this as major setback to their AI business goals, he quickly assembled a core team to develop a bold and realistic strategy. This MS team tested the security of DeepSeek R1 and found it legit. Satya immediately instructed his team to install DeepSeek R1 on Microsoft’s cloud and sell access to it to customers alongside products from OpenAI and Microsoft itself.

Microsoft Customers can now select from over 1,900 different models on Azure, including ones made by Meta and OpenAI, and other startups like as Cohere, Mistral, Stability AI and now DeepSeek. A model’s usage can cost a customer around $X through OpenAI or $0.1X via DeepSeek but Microsoft gets paid for the cloud computing, cybersecurity protections, data storage and other upsold services.

So to hedge its bets well in AI, MS has also been developing its own proprietary models called MAI, says Mustafa Suleyman, who heads the Copilot team of Microsoft. Like all companies, whether big, medium or small who are bewildered how AI will reshape their business, same applies to Microsoft too.

Microsoft's partnership with OpenAI began around Mid-2019 with its US $1 billion in OpenAI. In spite of this investment, like the rest of the world, Microsoft was also caught in surprise with release of ChatGPT in Nov 22. Since this, it has caught up in the AI world through release of the Copilot App. But ChatGPT still has above ~80% market share in the AI Chatbot market compared to Perplexity's ~10% and Copilot ~5% market share.

Microsoft Copilot continue to add global customers as Suleyman and his team keep adding impressive features and capabilities to the Copilot product. Users (including me) are now able to see substantive productive gains by using Copilot App including generating report from documents hundreds of pages long to neat one-pager cogently summarised.

GitHub Copilot surpassed 15 million users as of March 2025, and the Copilot Corporate version has hundreds of thousands of businesses as customers, with their engagement tripling over the past year. Microsoft now generates billions of $ in AI Revenue. 70% of global Fortune 500 companies now use Copilot.

MS CEO Satya Nadella himself uses Copilot extensively. As one use case, he loads his Podcast transcripts into the App so that he can chat on the go with the Copilot voice assistant regarding the podcast. Nadella appears not perturbed of DeepSeek undermining the value of Microsoft investment in OpenAI or the buildout of Microsoft data centers. His views are - more efficient AI will eventually mean a larger global need for AI services and data centre infrastructure.

Satya Nadella believes that AI will rapidly commoditize like what happened to coal in 18th/19th century and air travel in early 21st century. One of his favourite economic theories is the Jevons paradox. This theory predicts that as a resource becomes more and more available & efficient for mass usage, its consumption increases. Smartphones are also good example for this paradox.

As AI continues to evolve, Microsoft’s combination of collaboration, competition, and caution may prove to be a winning formula—not just for the company, but for the broader technology industry.

This vision is reflected in Microsoft’s product strategy, which emphasises integration of AI features across its entire software ecosystem. From AI-powered copilots in Office apps to advanced analytics in Azure, Microsoft aims to embed intelligence into every layer of its offerings. Nadella believes that the real value of AI lies in its ability to augment human capabilities and empower people to achieve more.

In closing, AI promises to unlock enormous amount of economic growth. But like for any new technologies, there are likely initial pains & scepticism too for AI implementation, adoption, acceptance and value creation. Similar to invention and adoption of steam engine, electricity, transportation, penicillin, computer and the internet, what follows after initial pains would be massive economic and productivity gains.