Google DeepMind Closes Three AI Deals in One Week to Supercharge Gemini

Google DeepMind closed three separate deals with AI startups in rapid succession in late January 2026, acquiring Common Sense Machines, striking a licensing agreement with Hume AI, and taking an equity stake in Tokyo-based Sakana AI. The moves signal a deliberate strategy to fill critical capability gaps in Google's flagship Gemini model.

Key Highlights

• Common Sense Machines acquired to boost 3D spatial reasoning in Gemini
• Hume AI licensed for emotional voice intelligence, with CEO and engineers joining DeepMind
• Sakana AI receives strategic investment for regional AI expansion in Japan
• All three deals closed within days of each other

Common Sense Machines: 3D Vision for Gemini

The Massachusetts-based startup, employing roughly 12 people, specialized in AI models that convert 2D images into structured 3D objects. Co-founder Tejas Kulkarni previously worked at Google DeepMind. The company was last valued at approximately $15 million after raising $10 million from investors including Andreessen Horowitz.

The acquisition brings technology that reduces hallucination in visual outputs and improves coherence across frames and viewpoints. Applications span robotics simulation, augmented reality, and video generation — areas where Gemini has been competing against OpenAI's Sora and other multimodal models.

Hume AI: Emotional Intelligence for Voice

Rather than a full acquisition, Google structured a licensing agreement with Hume AI, known for its models that recognize emotions in voice. CEO Alan Cowen and approximately seven engineers joined DeepMind, while Hume continues to operate independently with projected revenue of $100 million for the current year.

The technology analyzes tone, prosody, and emotional cues in speech, directly strengthening Gemini's voice interaction capabilities. This is particularly significant as voice interfaces become a primary way users interact with AI assistants across Google's product ecosystem.

Sakana AI: Regional AI Strategy in Japan

Google took an equity stake in Sakana AI, Japan's highest-valued AI startup at $2.5 billion. The company was founded in 2023 by David Ha, who formerly led Google Research in Japan, and Llion Jones, a co-author of Google's original Transformer architecture paper.

Sakana develops models inspired by collective intelligence and evolutionary systems. Its code agent notably ranked 21st among the top 1,000 human coders. The investment targets financial institutions and government agencies requiring strict data protection, combining Google's AI models with Sakana's "AI Scientist" and "ALE Agent" technologies.

A Calculated Strategy

The three deals reflect Google's approach of acquiring talent and licensing technology through narrowly scoped transactions rather than large-scale acquisitions that could trigger antitrust scrutiny. Each deal fills a specific gap: 3D spatial reasoning, emotional voice understanding, and regional market penetration.

As the AI race intensifies between Google, OpenAI, Anthropic, and Meta, these strategic moves position Gemini to compete more effectively across multiple modalities and geographies.

Source: The Decoder