AI and Valuation: Pricing Disruptive Technology

Introduction

Artificial intelligence is the most significant thematic driver of M&A valuations since the cloud computing wave of the 2010s. Approximately one-third of the 100 largest corporate M&A transactions in 2025 cited AI as part of the strategic rationale. In the technology sector specifically, nearly all of the largest deals referenced AI in their deal thesis. The average revenue multiple for AI-related M&A in 2025 reached 25.8x, dramatically above the 3-7x range for standard SaaS companies. Total tech M&A deal value surged 65% year-over-year in January 2026 alone, driven overwhelmingly by AI infrastructure transactions.

But AI's impact on valuation is more nuanced than a uniform premium. The market is developing a tiered framework that differentiates between AI companies based on the depth and defensibility of their technology, and applying very different multiples to each tier.

Tier 1: AI Infrastructure (Highest Multiples)

Companies building the foundational layers of AI: semiconductor designers (NVIDIA, AMD), cloud computing infrastructure (hyperscalers), data center operators (Equinix, Digital Realty), and large language model developers (OpenAI, Anthropic). These assets command the most extreme valuations because they are the "picks and shovels" of the AI revolution: every AI application depends on them, and supply is constrained.

Large language model vendors command revenue multiples of approximately 54.8x, reflecting both the transformative potential and the winner-take-most dynamics of foundation model development. Data intelligence companies trade at approximately 41.7x revenue. NVIDIA's market cap exceeded $3 trillion in 2025, making it one of the three most valuable companies globally.

Tier 2: AI-Native Applications (High Multiples)

Companies whose core product is built around AI: AI-powered cybersecurity platforms (like Wiz, acquired by Alphabet for $32 billion), AI-driven drug discovery (Recursion Pharmaceuticals, Isomorphic Labs), and AI-native enterprise software. These companies have AI embedded in their core architecture, not bolted on as a feature. They command 30-50% premiums over comparable non-AI software.

Tier 3: AI-Enhanced Products (Moderate Premium)

Established software and technology companies that have integrated AI capabilities into existing products (CRM with AI-powered recommendations, ERP with AI-driven forecasting, marketing platforms with AI content generation). The AI premium for Tier 3 companies is smaller (10-20% above non-AI peers) and increasingly under pressure as AI features become table stakes rather than differentiators.

AI Valuation Premium

The incremental valuation multiple that a company commands because of its AI capabilities, measured relative to comparable non-AI businesses. The premium reflects the market's assessment that AI creates sustainable competitive advantages through proprietary data assets, unique model architectures, network effects (more users generate more data, which improves the model), and high switching costs. The premium varies dramatically by tier: infrastructure-grade AI assets command 3-5x the multiple of comparable non-AI businesses, while feature-level AI enhancements may command only 10-20% premiums that erode as competitors adopt similar capabilities.

Revenue Quality Is Harder to Assess

Not all AI revenue is created equal. Consumption-based pricing (common for AI API services, where the customer pays per query or per token) creates variable revenue streams that are harder to project than traditional subscription models. A company reporting $100 million in AI-related revenue may have very different visibility depending on whether that revenue comes from annual contracts (predictable) or usage-based pricing (volatile).

The Moat Question Is Unresolved

The most consequential valuation debate in AI is whether foundation models will become commoditized (in which case the premium dissipates) or whether network effects and data advantages create durable moats (in which case the premium is justified). OpenAI's lead in general-purpose language models is significant, but open-source alternatives (Meta's Llama, Mistral) are closing the quality gap rapidly. If the model layer commoditizes, value accrues to the application layer (companies using AI to solve specific vertical problems) rather than the infrastructure layer.

These deals illustrate the range of AI-driven valuations: from pure infrastructure plays (NVIDIA, where the AI premium accounts for the majority of the valuation) to strategic acquisitions where AI is part of a broader thesis (Alphabet/Wiz, where AI-powered security was one dimension of a cloud platform strategy).

For investment bankers advising on AI-related transactions, the valuation approach combines standard tools with AI-specific adjustments:

Use [EV/Revenue](/guides/valuation-investment-banking/ev-revenue-pe-other-multiples-when-ev-ebitda-not-enough) as the primary multiple for high-growth AI companies, because most are pre-profit or have EBITDA margins compressed by growth investment. Segment the peer group into the appropriate tier (infrastructure vs. application vs. feature-level) to avoid comparing NVIDIA to a small AI SaaS startup.

Supplement with a [DCF](/guides/valuation-investment-banking/walk-me-through-dcf-end-to-end-framework) that models the path to profitability. AI companies investing heavily in model training, compute infrastructure, and talent will show near-term losses that give way to operating leverage as revenue scales. The projection period should be 7-10 years to capture this transition, and the terminal value should reflect mature, at-scale margins (which for software platforms can reach 30-40% EBITDA).

Assess the moat. The valuation should explicitly address whether the company's AI advantage is defensible. Proprietary training data that improves with use (a data flywheel), patented model architectures, and exclusive partnerships create lasting value. Generic AI features that any competitor can replicate with off-the-shelf models do not.

Data Flywheel (AI Moat)

A competitive advantage where a company's AI models improve as more users generate more data, which trains better models, which attract more users, creating a compounding cycle. Companies with strong data flywheels (Google Search, Tesla Autopilot, Amazon product recommendations) build AI advantages that are extremely difficult for competitors to replicate because the moat grows wider with use. In AI valuation, the strength of the data flywheel is one of the most important qualitative factors: a company with a proven flywheel (measurable model improvement tied to data volume) deserves a structurally higher multiple than one without, because the flywheel creates a self-reinforcing competitive advantage that persists even as the underlying AI technology commoditizes.