Google

Blackstone is investing $5 billion in a new AI infrastructure venture, partnering with Google. This venture will utilize Google’s specialized Tensor Processing Units (TPUs) to meet surging AI compute demand. The company aims to deploy 500 megawatts of capacity by 2027 and plans significant expansion, strengthening Google’s position against competitors like Nvidia in the AI hardware race. Blackstone is also increasing its AI investments with other firms.

Alphabet’s stock has surged 140%, driven by its strong cloud performance and strategic positioning in the AI boom. At Google I/O, investors anticipate concrete roadmaps for Gemini updates, AI agents, agentic shopping, and AI monetization. Google Cloud’s rapid growth and custom AI chip strategy are also key focuses. The partnership with Anthropic, while raising concentration questions, offers a hedge, ensuring Google benefits from AI infrastructure demand regardless of model choice.

Alphabet’s stock surged following strong Q1 results, boosted by cloud revenue growth. This propelled Google toward its best month since 2004. In contrast, Meta’s shares dropped sharply on news of increased AI spending, highlighting Wall Street’s mixed reaction to tech’s AI investments. While Alphabet sees AI success, Meta faces challenges in demonstrating returns on its substantial expenditures, leading to a neutral rating from JPMorgan.

Top AI researchers leaving tech giants like Meta and Google are launching new startups, attracting significant funding. These ventures focus on specialized AI innovations, often pursuing research neglected by larger companies. Investors are betting heavily on the expertise of these pioneers, fueling a rapid growth in early-stage AI labs and reshaping the industry landscape.

Google is investing up to $40 billion in AI research firm Anthropic, with an initial $10 billion cash infusion. This expands a prior collaboration, including securing 5 gigawatts of computing capacity. The deal deepens Google’s strategic alliance in the competitive AI landscape, following significant prior investments and Anthropic’s recent $5 billion Amazon partnership. This highlights a major industry race for AI dominance.

Google is bifurcating its Tensor Processing Units (TPUs) into specialized chips for AI training and inference, aiming to challenge Nvidia’s dominance. This move is driven by the rise of AI agents and the need for tailored performance. The new TPUs offer significant performance gains, with the inference chip incorporating more SRAM for efficient AI operations. This strategy aligns with industry trends of custom AI silicon development, with major tech players also investing heavily in in-house hardware.

Google is reportedly partnering with Marvell Technology to design two new AI chips, marking a shift in its silicon strategy previously dominated by Broadcom. This move, potentially including a specialized memory processing unit, could reshape the AI chip landscape. Marvell’s stock has surged on the news, while Broadcom saw a dip. This development highlights the growing demand for custom AI accelerators and the critical role of firms like Marvell and Broadcom in their design and production.

Google integrates its Gemini AI chatbot with the Nano Banana image generator, allowing users to create personalized AI artwork from their personal photo libraries. This feature, part of Google’s Personal Intelligence, enables users to generate images based on specific prompts and their own photos, enhancing user convenience and AI context.

Google is expanding its strategic alliance with Intel, committing to multiple generations of Xeon processors for its AI data centers. This move aims to bolster Intel’s position against competitors in the AI chip market, acknowledging the re-emerging importance of CPUs alongside accelerators. The partnership also includes ongoing collaboration on Infrastructure Processing Units (IPUs).

Meta launched Muse Spark, its first major AI model since acquiring Scale AI. Developed by Meta Superintelligence Labs, it focuses on speed and efficiency, excelling in scientific reasoning and problem-solving. The model aims to boost Meta’s AI presence against competitors like OpenAI and Google, and will be integrated into Meta’s platforms and smart glasses, with potential for paid API access and new revenue streams.