Salesforce Buys Cognition Labs for $2B

Salesforce has agreed to acquire Cognition Labs, the AI startup behind the autonomous coding agent Devin, in a deal valued at approximately $2 billion. The acquisition marks one of the largest AI startup deals of 2025 and signals Salesforce's aggressive push to embed autonomous AI agents across its enterprise software ecosystem.

The deal, which is expected to close in the coming months pending regulatory approval, will bring Cognition Labs' roughly 50-person team under Salesforce's AI division. It represents a significant premium over Cognition's previous valuation of around $2 billion during its last funding round, underscoring how fiercely enterprise software giants are competing for top-tier AI talent and technology.

Key Facts at a Glance

• Deal value: Approximately $2 billion, making it Salesforce's largest AI-focused acquisition to date
• Target: Cognition Labs, creator of Devin — widely recognized as the first fully autonomous AI software engineer
• Strategic rationale: Bolsters Salesforce's Agentforce platform with advanced autonomous coding and task-completion capabilities
• Team size: Cognition Labs employs roughly 50 engineers, many from elite competitive programming backgrounds
• Timeline: Expected to close within Q3 2025, subject to standard regulatory review
• Funding history: Cognition Labs previously raised over $170 million from investors including Founders Fund, Khosla Ventures, and others

Why Cognition Labs Matters to Salesforce

Cognition Labs burst onto the AI scene in early 2024 with Devin, a product that stunned the developer community by demonstrating the ability to autonomously write, debug, and deploy code. Unlike conventional coding assistants such as GitHub Copilot or Cursor, which function as sophisticated autocomplete tools, Devin operates as a fully autonomous agent capable of handling complex, multi-step engineering tasks with minimal human oversight.

The startup was founded by Scott Wu, a former competitive programming champion, along with a team of engineers with deep expertise in reasoning and planning systems. Their core technology focuses on building AI systems that can break down complex tasks into subtasks, maintain long-term context, and self-correct — capabilities that extend far beyond code generation.

For Salesforce, this acquisition is not merely about adding a coding tool. It is about acquiring the underlying agentic AI architecture that powers autonomous task execution. This technology can be adapted to customer service workflows, sales automation, data analysis, and virtually any business process that requires multi-step reasoning.

Agentforce Gets a Major Upgrade

Salesforce has been building its Agentforce platform as the centerpiece of its AI strategy since late 2024. The platform allows businesses to deploy autonomous AI agents that can handle customer inquiries, process transactions, and manage workflows without constant human supervision. CEO Marc Benioff has repeatedly described Agentforce as 'the future of enterprise AI.'

However, Agentforce has faced criticism from analysts who argue that its agents lack the deep reasoning capabilities needed to handle truly complex, multi-step tasks. Cognition Labs' technology directly addresses this gap. By integrating Devin's planning and execution engine into Agentforce, Salesforce can offer agents that don't just respond to prompts but actively solve problems.

This positions Salesforce more competitively against rivals like Microsoft, which has been integrating AI agents through its Copilot Studio platform, and ServiceNow, which has invested heavily in workflow automation powered by large language models.

The Broader AI Acquisition Arms Race

Salesforce's move comes amid a wave of AI acquisitions and talent deals reshaping the enterprise software landscape. The deal follows a pattern of major tech companies snapping up AI startups to accelerate their product roadmaps rather than building capabilities in-house from scratch.

Recent comparable deals include:

• Amazon's $4 billion investment in Anthropic, securing preferential access to Claude models
• Microsoft's multi-billion-dollar partnership with OpenAI, now deeply integrated into Azure and Office 365
• Google's acquisition of DeepMind and continued investment in AI-native startups
• Databricks' $1.3 billion acquisition of MosaicML for custom model training
• Snowflake's purchase of Neeva to integrate AI-powered search into its data platform

What makes the Cognition Labs deal distinctive is the focus on agentic capabilities rather than foundational models. Salesforce is not buying a large language model — it already partners with OpenAI, Anthropic, and Google for foundational model access. Instead, it is acquiring the orchestration layer that turns raw model intelligence into autonomous action.

This reflects a maturing understanding in the industry: the value in AI is shifting from model training to agent orchestration. Companies that can build reliable, safe, and effective autonomous agents will capture disproportionate value in the enterprise market.

What This Means for Developers and Businesses

The acquisition carries significant implications for multiple stakeholder groups. For enterprise customers already using Salesforce, the integration of Cognition's technology could unlock powerful new capabilities within existing workflows.

For Salesforce customers:
- Expect autonomous agents capable of handling complex multi-step business processes, from invoice reconciliation to customer onboarding
- Agentforce agents will likely gain the ability to write and modify custom Apex code, Salesforce flows, and integrations without requiring dedicated developers
- Reduced dependency on external development teams for platform customization

For developers:
- Cognition's technology could democratize Salesforce development, lowering the barrier to building on the platform
- Professional Salesforce developers may need to adapt, shifting toward oversight and governance roles rather than hands-on coding
- The Devin product may continue as a standalone offering, though its roadmap will likely prioritize Salesforce ecosystem integrations

For the broader AI industry:
- Validates the agentic AI thesis — autonomous agents that act, not just advise, represent the next frontier
- Sets a valuation benchmark for other agentic AI startups considering exits or fundraising
- Increases pressure on competitors like Microsoft and Google to make similar acquisitions or accelerate internal development

Challenges and Risks Ahead

Despite the strategic logic, the deal is not without risks. Integrating a fast-moving startup into a $30 billion enterprise software company is notoriously difficult. Salesforce's track record with acquisitions is mixed — its $27.7 billion purchase of Slack in 2021 has delivered slower-than-expected synergies, and some critics argue the messaging platform has lost momentum under Salesforce's ownership.

There are also technical challenges. Devin has impressed in demonstrations but has faced scrutiny over its real-world reliability. Some independent benchmarks have shown that autonomous coding agents still struggle with production-grade codebases, edge cases, and security-sensitive tasks. Scaling this technology to meet enterprise-grade reliability standards will require significant engineering investment.

Additionally, the deal raises questions about AI governance. Autonomous agents that can write and execute code within business-critical systems introduce new categories of risk. Salesforce will need to build robust guardrails, audit trails, and human-in-the-loop controls to satisfy enterprise security and compliance requirements.

Looking Ahead: The Race for Autonomous Enterprise AI

Salesforce's acquisition of Cognition Labs is more than a headline-grabbing deal — it represents a strategic bet on the future of enterprise software. The company is wagering that within the next 2 to 3 years, autonomous AI agents will handle a significant portion of the work currently performed by human knowledge workers and developers.

If the integration succeeds, Salesforce could establish a meaningful lead in the agentic enterprise AI category, differentiating itself from competitors who remain focused on copilot-style assistants. The company has signaled plans to showcase initial integrations at its Dreamforce conference later this year, where it is expected to demonstrate Cognition-powered agents operating within the Salesforce platform.

For the AI startup ecosystem, this deal sends a clear message: the acqui-hire era is giving way to full-scale acquisitions of companies with proven autonomous AI capabilities. Startups building agentic systems — whether for coding, customer support, finance, or operations — now have a compelling proof point for both their technology thesis and their exit potential.

The $2 billion question is whether Salesforce can turn Cognition's impressive technology into a reliable, scalable enterprise product. The answer will shape not only Salesforce's trajectory but the broader direction of how businesses adopt and deploy autonomous AI.