Mid-market artificial intelligence M&A is growing in value despite a fall in volume
Mid-market artificial intelligence M&A is facing a structural paradox: the value of deals involving AI targets rose by 127 per cent year-on-year in the first half of 2025, whilst the volume fell by around 20 per cent. The conclusion is clear, as set out in the latest report by Baker Tilly’s technology M&A advisers: buyers are paying premiums for capability and talent, not for revenue.
The segment analysed comprises AI companies with an enterprise value of less than $100 million, funded from pre-seed stages through to Series B rounds of under $50 million. It is a broad group, with limited reported revenue and low capital intensity, divided between defensible ‘AI-native’ vertical businesses and a long tail of fragile wrappers. Excluded from the analysis are cutting-edge model labs, hyperscaler infrastructure and silicon, which are out of reach at this scale.
For a founder considering selling, raising a round of funding or simply understanding where their company stands on this map, the report sets out six market indicators and a valuation framework that are worth reviewing before making any decisions.
Six trends that are redefining the valuation of mid-market AI
Parity of open-weight models
Models such as Llama 4, Qwen 3.5/3.6, DeepSeek V4 and GLM-5 match closed-source models in practical workloads and drastically reduce the cost of deployment. Self-hosting a 70B-class model costs between $8,000 and $12,000 a month, compared with over $45,000 in API fees. This provides a margin boost for those who adopt it, but also erodes the competitive moat for those who based their competitive advantage on access to the model.
Fall in API prices and squeeze on margins
Prices for LLM APIs fell by around 80 per cent in a year. Each price cut squeezes the margins of wrapper providers, whilst leaving the business model of vertical agents – which set prices based on results rather than usage – intact. AI-native applications operate with gross margins of 50–70 per cent, still below the 75–85 per cent seen in traditional SaaS, but with a track record of improvement that the market is beginning to reward.
Concentration of agency
More than half of Y Combinator 's spring 2025 cohort built agents. Vertical agents accounted for around 55 per cent of M&A deals in the sector and 57 per cent of the capital invested in agents. At the same time, the formation of new general-purpose agent start-ups has come to a screeching halt: the seed funding window for general-purpose agents is closing, whilst niche vertical agents continue to raise capital with ease.
The SMB as a driver of demand
The adoption of AI among US SMEs with 10 to 100 employees jumped from 47 per cent to 68 per cent in 2025. No-code platforms and agent marketplaces (Salesforce AgentExchange, Claude Marketplace, GPT Store) are reshaping distribution and making access to SMB buyers more affordable, thereby expanding the pool of companies that can gain traction without a traditional sales force.
Regulatory divergence in Europe
The European AI Regulation turns regulatory compliance into a barrier to entry and, at the same time, a moat for the incumbent, which is already adapted to the new rules. The high-risk obligations, deferred until December 2027 by the Digital Omnibus, entail initial costs of between 200,000 and 500,000 euros. This makes operations more expensive for early-stage start-ups, but enhances the value of European businesses that are already compliant in the eyes of cross-border buyers seeking to reduce their own regulatory risk.
The AI-versus-wrapper’ line is already a line of assessment
Customer retention determines the valuation multiple. AI-native plans costing under $50 a month retain 32 per cent of customers, compared with 85 per cent for plans costing over $250. A cheap wrapper is equivalent to anacqui-hire; a vertical tool embedded in the workflow, with a Net Revenue Retention of over 110 per cent, becomes a bolt-on with a premium. Distinguishing between these two profiles is, according to the report, the most decisive due diligence exercise in the sector.
Here’s an article from our M&A Academy in which we discuss the 10 key factors for successful due diligence in the technology sector.
How M&A deals under $100 million work
The mid-market artificial intelligence M&A market is structurally opaque: only 21 per cent of the approximately 5,700 AI acquisitions recorded between 2020 and 2025 disclosed their transaction value. Even so, the underlying dynamics are clear: activity is driven by technical capability and talent, not revenue volume.
The predominant type of transaction combines two models. On the one hand, the ‘bolt-on’ acquisition of niche specialists, with MongoDB’s acquisition of Voyage AI (US$220 million) serving as a prime example. On the other hand, the acqui-hire of wrapper teams, typically valued at between US$1 million and US$3 million per engineer brought on board.
The market context supports this activity. In 2025, there were 2,698 SaaS M&A transactions – a record figure representing a 28 per cent increase – and 72 per cent of those targets already highlighted AI capabilities as part of their value proposition. Strategic buyers accounted for 62 per cent of mid-market deals, whilst private equity entered 2026 with some $3.7 trillion in dry powder – capital that is fuelling the appetite for buy-and-build strategies at the smaller end of the market.
The 2025–2026 valuation multiple correction is widening the gap between different profiles. The SaaS Capital index fell from around 7x ARR to around 3.8x, with approximately one trillion dollars in SaaS market capitalisation wiped out in a single quarter. Against the backdrop of this widespread correction, premium AI-native assets continue to trade at between 15x and 30x revenue, whilst software that is becoming commoditised is converging towards the lower multiples typical of traditional SaaS. The central question in any M&A process remains the same: is this company AI-native, or is it merely a wrapper that is defensible only for as long as its price advantage lasts?
If you’d like to find out more about the trends that are revolutionising the SaaS market and learn about its growth forecasts, please access the 2026 SaaS sector report compiled by our consultants specialising in technology M&A.
AI market growth forecast up to 2030
Rather than a single growth rate, the report presents scale as a spectrum of sensitivity, with a baseline scenario for population growth in the mid-to-high twenties in percentage terms. By segment, the projected compound annual growth rates (all of which are forward-looking estimates) are as follows:
- AI agents (agent-based): the most aggressive segment, with a reported CAGR of 46.3% through to 2030, although this should be treated as subject to change as capital becomes concentrated and the market matures.
- MLOps and data tools: a CAGR of 37.4 per cent, rising from $1.7 billion in 2024 to $39 billion in 2034, with a particular focus on vector databases, model evaluation and observability.
- Data labelling and cleansing: from between 4,000 and 5,000 million dollars in 2025 to around 17,000 million in 2030, at an approximate CAGR of 28 per cent.
- Vertical SaaS / vertical AI, the core of the mid-market segment: a CAGR of 11.5 per cent on a conservative basis, reaching a market worth $164,000 million by 2026, with sub-segments tracked by venture capital growing at rates of between 16 per cent and 23 per cent.
The driving force behind these figures is what the report describes as ‘vertical AI devouring horizontal SaaS’: AI-native vertical companies reach $100 million in ARR faster than any previous generation of SaaS. Their combination of ownership of the customer’s workflow and proprietary data creates a moat that a frontier model upgrade cannot easily bridge. That defensibility is, ultimately, what makes a vertical player investable and acquirable, whereas a horizontal wrapper is not.
What does this report mean if you’re considering selling or investing?
The message of the report for the founder of a mid-market AI company is clear: the market no longer rewards the mere fact of ‘having AI’ in the product; rather, it rewards evidence of defensibility — retention, NRR, ownership of the workflow — as opposed to reliance on an external model. That distinction is also the first thing any strategic buyer or private equity fund looks at during a due diligence process.
At Baker Tilly Tech M&A, we analyse in detail how your company is positioned on this valuation map, whether you are considering selling your tech company or exploring an acquisition within the sector. If you would like to find out more about how AI is driving valuation multiples in the Spanish tech mid-market, you can also consult our strategic guide on AI and valuation multiples in Tech M&A.
Market Research
Market Research on the AI Sector in the Mid-Market in 2026
Download the full report here and stay informed about the financial status and latest news of the most important companies in the market.
Frequently asked questions about M&A in the mid-market AI sector
An AI-native company builds ownership of the workflow and customer data, with retention rates exceeding 85 per cent on higher-priced plans and an NRR above 110 per cent. A wrapper relies on access to a third-party model, retains around 32 per cent on budget plans, and its sales value is closer to that of an ‘acqui-hire’ of talent than to that of a recurring business.
This is because buyers are paying premiums for technical expertise and specialised equipment, not for recurring revenue. In the first half of 2025, the value of deals involving AI targets rose by 127 per cent year-on-year, despite transaction volumes falling by around 20 per cent.
Premium AI-native assets are trading at between 15x and 30x revenue, well above commoditised software, which is converging towards 3.8x ARR on the SaaS Capital index following the 2025–2026 revaluation. The difference depends almost entirely on whether the buyer perceives structural defensibility or dependence on an external model.
High-risk bonds, maturing in December 2027, entail initial costs of between 200,000 and 500,000 euros. This penalises early-stage start-ups that have not yet invested in compliance, but enhances the value of European companies that are already compliant in the eyes of cross-border buyers seeking to reduce their regulatory exposure.
AI agents lead the way with a projected CAGR of 46.3% through to 2030, followed by MLOps and data tools at 37.4%. Both figures should be treated as estimates subject to market maturation and the gradual concentration of capital amongst established vertical players.
