Google To Integrate Ads Into AI-Powered Search Overviews

Google has announced plans to incorporate search and shopping ads within its AI-generated answers, marking a significant expansion of its advertising capabilities. This initiative, which will be tested in the United States, follows the introduction of the AI Overviews feature at Google’s recent I/O conference. The ads will appear in a ‘sponsored’ section, tailored to the relevance of the user’s query.

Strategic Expansion in AI and Advertising

This move underscores Google’s strategy to leverage its dominance in traditional search advertising by integrating advanced generative AI technologies. The initiative aims to boost ad sales, a major revenue source, which saw a 13% increase to $61.7 billion in Q1 2024. By embedding ads within AI-generated search results, Google seeks to maintain its competitive edge and revenue growth amidst evolving digital landscapes.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

Ongoing Developments and Future Directions

Google will continue refining new ad formats, drawing on feedback from advertisers. Enhancements showcased at the I/O conference, including updates to the Gemini chatbot and search engine improvements, highlight Google’s commitment to advancing AI across its services.

Google’s integration of ads into AI-driven search overviews represents a forward-thinking approach to digital advertising. As the company navigates the intersection of AI innovation and commercial strategy, these developments are set to influence the broader advertising ecosystem significantly.

The Future Forbes Realty Global Properties

Meta Unveils 816 Billion-Parameter AI Model For Mac, But It Requires 1TB Of RAM

by THEFUTURE.TEAM

October 16, 2026

Meta has introduced what it describes as the most capable open-weight AI model available in the United States, making a strategic bet on Apple’s Mac ecosystem. The release reflects the company’s ambition to strengthen its position in open AI while bringing advanced models into professional development and creative workflows.

A Powerful Model With High Hardware Requirements

Called Muse Spark Open, the model has 816 billion parameters and a context window of 1 million tokens. Meta said benchmark results published on Thursday show it outperforming other widely used open-weight systems across key evaluations.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

Natively multimodal, Muse Spark Open is designed for complex coding and agentic tasks, extending beyond conventional chatbot use. However, local deployment requires substantial computing power. According to Engadget, the model needs at least 1TB of RAM, with the best performance expected on a Mac Studio equipped with Apple’s M5 Ultra chip. These requirements may limit local use to developers and organisations with access to high-end hardware.

Built For Real-World Development

Meta has paired Muse Spark Open with Muse Code, its AI coding terminal, and tested it across agentic workflows. In one project, the model helped develop a Mastodon client, handling bug fixes, new features and the build process.

The company also deployed the model in DaVinci Resolve, the professional video-editing platform, where it reviewed and cropped footage. These examples demonstrate Meta’s aim to position open-weight AI as a tool for practical production work, from software development to creative tasks, rather than just research demonstrations.

The Strategic Case For Open-Weight AI

Open-weight models give developers and organisations greater flexibility to customise AI, integrate it into internal tools and build specialised workflows. Depending on deployment, they can also reduce costs and provide more control over infrastructure than relying entirely on proprietary services.

However, access to model weights does not eliminate the cost of running advanced AI. Muse Spark Open’s reported memory requirements could make API access a more practical option for many developers than local deployment.

Expanding Access Beyond Mac Hardware

Meta is making Muse Spark Open available through multiple API platforms, allowing developers to use the model without running it on their own machines. This broadens access beyond users with high-end Mac hardware while retaining the option of local deployment for those able to support it.

The release combines Meta’s push to strengthen its position in the open-weight AI ecosystem with a focus on professional users. Wider adoption will depend on the model’s performance, practical usefulness and the resources required to run it.

Breaking news