Andreessen Horowitz is doubling down on AI applications beyond coding and infrastructure, backing a litigation technology startup while the underlying economics of AI models are shifting dramatically. The venture capital firm participated in a $3.8 million seed round for Concorda, a platform that combines legal research, matter management, and document drafting for law firms, signaling a16z's continued bet on vertical AI applications even as foundational model costs collapse.

What Is Concorda and Why Does a16z Care?

Concorda announced its seed funding round on September 8, 2026, with the capital led by Sazze Partners and supported by a16z's speedrun startup accelerator program, alongside investors including Amino Ventures, Gaingels, and G2C Ventures. The platform is designed to guide legal professionals through the entire litigation lifecycle, from client intake through settlement or trial, using generative artificial intelligence (AI) and traditional software automation to streamline repetitive tasks.

The investment reflects a broader trend in venture capital: while headline-grabbing AI infrastructure deals dominate the news, a16z and other major firms are quietly backing specialized AI tools tailored to specific industries. Legal technology represents a particularly attractive market because law firms have high hourly billing rates, making even modest efficiency gains valuable enough to justify software spending.

How Are AI Model Costs Reshaping the Economics of AI Startups?

The timing of Concorda's funding coincides with a seismic shift in AI economics. According to data highlighted by Andreessen Horowitz, citing Goldman Sachs research, the cost of using large language models (LLMs), which are AI systems trained on vast amounts of text to generate human-like responses, has fallen approximately 95% in just three years. The index measuring average price per million tokens, the basic units of text that AI models process, dropped from 100 to approximately 5 between March 2023 and September 2026.

To put this in perspective, personal computer prices required roughly 15 years to achieve a similar decline during the technology boom of the 1980s and 1990s. The difference becomes even more pronounced when factoring in improvements to model performance. A quality-adjusted measure comparing model prices with intelligence capabilities fell to nearly zero over the same three-year period, meaning AI models are not only cheaper but also dramatically smarter per dollar spent.

This rapid deflation in AI costs stems from several converging factors:

• More Efficient Hardware: Specialized AI chips have become more powerful and cost-effective, reducing the computational expense of running models.
• Increased Competition: Multiple companies developing large language models have intensified price competition, pushing down costs across the industry.
• Smaller Specialized Models: Developers are creating smaller, task-specific models that require less computing power than massive general-purpose systems.
• Training and Inference Improvements: Better algorithms for training models and running them in production have reduced overhead and resource requirements.

What Does This Mean for AI Startups Like Concorda?

The collapse in model costs creates both opportunity and pressure for startups building AI applications. On the positive side, lower costs make it economically feasible to build specialized tools for niche markets like litigation technology. Concorda can now afford to offer sophisticated AI-powered features without facing prohibitive infrastructure expenses.

However, the same cost collapse that enables startups also intensifies competition. As basic model access becomes increasingly commoditized, AI providers must generate revenue through higher usage volumes, premium services, and enterprise-focused products rather than simply reselling access to expensive models. This dynamic explains why a16z is backing vertical applications like Concorda; the real value creation increasingly lies in domain expertise and user experience, not in owning the underlying AI model.

Concorda's approach reflects this reality. The platform doesn't attempt to build its own large language model from scratch. Instead, it leverages existing generative AI capabilities and combines them with legal domain knowledge, matter management workflows, and drafting automation tailored specifically to litigation professionals. This strategy allows the startup to move quickly and focus capital on product development and customer acquisition rather than burning cash on model training.

How Are Other AI Investors Responding to These Market Dynamics?

The broader venture capital market is signaling that the era of winner-take-all AI dominance may be ending. Cognition, a startup developing the Devin coding assistant, recently raised $2 billion at a $48 billion valuation, led by Andreessen Horowitz alongside Accel, Founders Fund, General Catalyst, and Avenir. The round, which closed just four months after Cognition's previous fundraise at a $26 billion valuation, demonstrates that investors still see room for multiple major players in AI coding, one of the technology's most significant applications.

Notably, a16z led this round in Cognition despite having previously backed Cursor, another coding assistant that sold to SpaceX for $60 billion in April 2026. The fact that a16z is willing to back a Cursor competitor signals confidence that the AI coding market is large enough to support multiple winners, rather than consolidating around a single dominant player. This contrasts sharply with earlier venture capital cycles, where winner-take-all dynamics often emerged in software markets.

The investments in both Concorda and Cognition reflect a16z's conviction that as foundational AI models become cheaper and more commoditized, the value creation shifts to specialized applications and domain expertise. Litigation technology and coding assistance represent two distinct verticals where AI can deliver meaningful productivity gains, but success requires deep understanding of the specific workflows and pain points in each industry.

What Should Law Firms and Other Industries Expect?

The combination of falling AI model costs and increased venture capital backing for vertical applications suggests that AI adoption across professional services will accelerate significantly. Law firms, accounting firms, consulting practices, and other knowledge-intensive businesses can expect an influx of specialized AI tools designed specifically for their workflows. These tools will become increasingly affordable as underlying model costs continue to decline.

However, the real competitive advantage will accrue to startups and established software vendors that can combine AI capabilities with deep domain expertise. Concorda's focus on the litigation lifecycle, from intake through trial, exemplifies this approach. Rather than offering a generic AI assistant, the platform is purpose-built for legal professionals, incorporating workflows and terminology specific to litigation practice.

For investors and entrepreneurs, the lesson is clear: the AI infrastructure layer is becoming a commodity, but the application layer remains wide open. Startups that can identify specific industries or workflows where AI can deliver outsized productivity gains, and that can build products tailored to those use cases, will attract venture capital and build defensible businesses. a16z's participation in Concorda's seed round signals that this thesis is gaining traction among top-tier venture investors.