The rise of artificial intelligence is gradually transforming corporate finance, and valuation is no exception. Since 2023, professional publications - including the work of Grbenic & Jagrič (EBVM 2025) and the positions of the IVSC - have shown a clear trend: AI is becoming a powerful tool for analyzing, predicting and automating certain tasks, but it does not replace human expertise, particularly when value depends on strategic, qualitative or legally sensitive elements.
This article sets out what algorithms can really predict in valuation, what they don't yet know, and the practical consequences for the professional valuations carried out by XVAL.
1. What AI can predict in business valuation
The studies cited in the EBVM (2025-03 edition), as well as analyses by the IVSC in 2025, show that machine learning algorithms provide real added value in three main areas.
1.1. Statistical prediction of company values
Supervised machine learning models - such as penalized regressions, random forests, gradient boosting or certain neural networks - can statistically predict the value of a company by exploiting a large volume of financial, sectoral or macroeconomic variables.
According to the article "Artificial Intelligence in Business Valuation - Part I", these models often achieve better predictive performance than conventional models when it comes to estimating a statistical value or deducing a likely valuation area.
XVAL can use these models to enrich quantitative analysis, refine value ranges and reinforce the consistency of comparables.
1.2. Automatic selection of comparables
Modern algorithms are effective at identifying truly similar companies and detecting those that deviate statistically from the sector. They can analyze several thousand companies simultaneously, which is beyond human capacity.
EBVM 2025 points out that AI significantly improves the selection of comparables, a central element in valuation by multiples.
1.3. Extraction and automated processing of financial data
Automatic Document Processing (ADP) models can quickly extract relevant information from financial reports, notes and PDF documents.
The work of Grbenic & Jagrič indicates that these technologies enhance the quality and speed of financial analysis prior to valuation.
2. What AI doesn't yet know, and what remains decisive for reliable valuation
Despite its advances, AI has structural limitations. The IVSC, in its document "Navigating the Rise of Artificial Intelligence in Valuation" (2025), stresses the need to maintain the valuator's professional responsibility.
2.1. The inability to integrate professional judgment
An algorithm cannot decide on the relevance of a balance sheet restatement, the inclusion of an unobservable risk in the data, or the impact of a strategic decision.
These trade-offs are the exclusive province of human expertise. Predictive models are no substitute for understanding the business model, analyzing management or assessing governance.
2.2. The impossibility of understanding future strategy
AI uses historical data. It cannot anticipate :
- a change of direction,
- a technological breakthrough,
- the launch of a new range,
- a major regulatory change,
- internal restructuring.
Strategic projections, the quality of the business plan and the intentions of management are essential in a DCF. This information is only accessible to the expert.
2.3. The inability to manage atypical contexts
Some valuations are "off-market": divorce, disputes between partners, transfer taxes, succession, restructuring. No automatic model can integrate these contextual, legal or human factors.
XVAL intervenes precisely in these complex valuations, where contextual interpretation is central.
2.4. AI does not choose methods
A predictive model can calculate, but not decide on, the appropriate valuation method: DCF, multiples, comparable transactions, asset valuation, relief-from-royalty, etc.
The choice depends on the nature of the company, its risks, its sector and its valuation objective. This decision involves reasoning, not calculation.
3. What this means for XVAL
The conclusion of EBVM publications and IVSC positions is clear: AI-assisted valuation is more accurate, but remains an expert process. To ensure quality, XVAL articulates its methods in three levels:
- AI-enhanced data extraction and analysis for more reliable calculation bases.
- Predictive models to improve statistical accuracy and refine comparables.
- Expert human intervention for interpretation, methodological arbitration, contextual consideration and determination of the final value.
AI is a tool. The expert remains the decision-maker.
Conclusion "AI and business valuation
Recent technical publications (EBVM, IVSC, EACVA) converge on one point: AI improves valuation, but does not replace it. It enhances statistical accuracy, optimizes comparables and speeds up data analysis. On the other hand, it can neither understand strategy, nor interpret legal issues, nor integrate the human elements that influence a company's value.
At XVAL, artificial intelligence enhances the quality of our analyses, but the final value is always based on human expertise, professional judgment and detailed understanding of context.
Request an appointment with a XVAL consultant for a valuation :
