Google and Microsoft flex their muscles in AI

BarcelonaAs the competition to dominate the future of artificial intelligence (AI) intensifies daily, Google and Microsoft held their respective developer conferences this week, Google I/O 2025 and Microsoft Build 2025, with AI taking center stage. Both digital giants showed their cards in a game where billions of euros and global technological leadership are at stake.

Google I/O 2025: A Display of Technological Power

The Shoreline Amphitheater in Mountain View, California, is the stage where Google has unleashed its full potential in AI. During the inaugural presentation, CEO Sundar Pichai didn't miss the opportunity to boast about the Nobel Prize won last year by Demis Hassabis, which other platforms receive. Pichai stated that Google's AI models occupy the top three spots in the world in terms of usage volume, thanks to a 50-fold increase in just one year. ChatGPT for searching for information, Google's web search engine incorporates AI Mode, a new tab designed to handle more complex queries than traditional ones, such as comparing multiple devices or finding the cheapest tickets for a game or concert. The feature is already available to users in the US. Google's AI models are designed to improve the user's personal style based on their previous correspondence, while Google Meet will offer real-time automatic translation of conversations, initially between English and Spanish. The Chrome browser will also be able to interrogate Google's AI directly, without having to install any extensions. This is especially true for those who are looking for AI models that are specifically designed to improve the user's personal style based on their previous correspondence. Google has also shown off new content generation models: Veo 3, capable of creating videos with realistic audio, visual effects and dialogue, and Image 4, which improves image generation. Perhaps to compensate, Google has also introduced the SynthID Detector tool, to verify

In the infrastructure field, the company has unveiled Ironwood, the seventh generation of its data center processor, specifically designed to power reasoning and large-scale AI workloads. According to Google, it offers 10 times the performance of US generations. It incorporates additional protections, adapting to a stricter regulatory framework, in contrast to the more permissive North American version. the user to payment with Google Pay. The function, still experimental, demonstrates both Google's ability to offer a complete experience within its ecosystem and the worrying possibility of contaminating the algorithms' responses with commercial interests. hardwareGoogle has surprised everyone with a revival of its legendary Glass augmented reality glasses. They now have a more conventional format, based on the company's Gemini AI and the Android XR platform, which has committed to adoption from brands like Samsung, Warby Parker, and Gentle Monster, similar to what Meta has done with Ray-Ban.

Also unveiled at I/O was Google Beam, a 3D video conferencing system that transforms traditional 2D video calls into immersive, realistic 3D experiences.

Microsoft Build 2025: AI for Developers

Meanwhile, in Seattle, Microsoft focused its Build 2025 conference on "AI agents" and the construction of what they've called "the open agentic web"—agentic AI is a type of artificial intelligence that focuses on autonomous systems that can make decisions and perform tasks without human intervention. One of the new developments is the evolution of GitHub Copilot, Microsoft's programming assistant, which already has 15 million users. The company has announced that GitHub Copilot is moving from being an assistant within the editor to an AI partner with agentic capabilities, which can act as an additional member of the development team, refactoring code, improving test coverage, fixing defects, and implementing new features autonomously (gulp!) We'll see where it all leads.

Rivaling DeepSeek and Meta's Llama, Microsoft has also announced that it will open source the GitHub Copilot Chat code, which will thus become part of the general open source repository of the world's most popular development tool. In the same vein of making it easier for app and service creators to adopt AI models in their products, the Azure AI Foundry and Windows AI Foundry platforms have also been presented. In addition, the company has unveiled Microsoft Discovery, a platform designed to accelerate scientific research using AI. According to Microsoft, it has allowed them to discover an immersion coolant for data centers in just 200 hours, a process that would otherwise have taken years.

However, perhaps the most significant thing about Build 2025 is Microsoft's distancing itself from OpenAI, despite being its main investor. The company has announced that the AI model repository on its Azure cloud will also incorporate Grok 3 models from xAI, the company founded by Elon Musk. Whether in response to tensions in its relationship with OpenAI or simply to offer more options to customers, it's clear that Microsoft is diversifying its sources of AI models.

Competitive Advantage

The Google and Microsoft conferences make it clear that both companies have a significant advantage over new competitors in the field of AI: they can integrate it into the products and services we already use every day, making them more efficient and useful.

This is no small advantage, because it allows both companies to capitalize on their enormous investments in AI through premium subscriptions to the same services that consumers and businesses already know. It's no coincidence that both have announced more expensive paid plans (such as Google AI Ultra for $249 per month) to access the most advanced capabilities of their models.

As the race for the most powerful AI continues at full speed, Google and Microsoft have made it clear that their goal is not only to lead in technology, but also in the ability to convert it into tangible revenue. Users receive more features, but we will also have to pay more to access them.