How Artificial Intelligence Is Impacting Software Valuations in 2026

Artificial intelligence is now one of the first topics buyers, investors, lenders, founders, boards, and appraisers discuss when a software company is being valued. That does not mean AI automatically increases value. It means AI has become a business valuation question: does the technology change cash flow, risk, comparability, or transferable asset value?

For a private software company, the answer can be positive, negative, or neutral. AI may support a stronger valuation when it improves customer outcomes, increases retention, creates pricing power, lowers recurring support costs, strengthens a proprietary data advantage, or expands the useful life of the company’s product. AI may reduce value when it makes a product easier to replace, compresses pricing, increases model and cloud costs, creates vendor dependence, weakens gross margin, introduces intellectual-property uncertainty, or causes customers and buyers to distrust unsupported marketing claims.

The practical lesson is simple: AI is not a valuation method. The core valuation methods still include the income approach, the market approach, and the asset approach, supported by professional judgment, documentation, and reconciliation. A discounted cash flow model still depends on revenue, margins, reinvestment, working capital, taxes, terminal value, and risk. EBITDA still must be normalized carefully. Market evidence still must be comparable. Source code, data, models, contracts, customer relationships, and legal rights still matter. A professional business appraisal should translate AI into those valuation inputs instead of treating the phrase “AI-powered” as a conclusion.

This article explains how artificial intelligence is affecting software valuations in 2026, with a focus on private software and SaaS companies. It is written for owners, CFOs, investors, attorneys, CPAs, lenders, and advisers who need a defensible framework rather than a hype cycle. It is educational only and is not legal, tax, accounting, audit, cybersecurity, securities, investment banking, lending, intellectual-property, or valuation advice for a specific company. Legal, tax, accounting, IP, cybersecurity, and regulatory conclusions should be confirmed with qualified professionals.

Quick AI Valuation Impact Table

The first step is to classify how AI affects the company’s actual economics. The following table uses hypothetical profiles and should not be read as a valuation conclusion or a market multiple guide.

Hypothetical software profile	How AI affects the business	Valuation direction is most likely to depend on	Discounted cash flow focus	Market approach focus	Asset approach focus
AI-native vertical workflow platform	AI automates a high-value industry workflow with proprietary domain data	Retention, pricing power, workflow depth, data rights, model cost, and governance	ARR growth, churn, gross margin, R&D, terminal risk	Whether guideline companies share workflow depth, data advantage, and margin quality	Developed technology, data rights, IP, source code, and customer relationships
AI-enhanced incumbent SaaS	AI features improve an existing product, but the core model remains SaaS	Whether AI increases expansion, retention, or margins after required investment	Upsell, support cost, model hosting cost, sales efficiency	Compare to similar mature SaaS firms only after quality adjustments	Capitalized software, source code, customer relationships, and AI feature assets
Horizontal tool exposed to agentic substitution	AI agents may perform tasks previously handled inside the product	Substitution risk, switching costs, product roadmap credibility, and customer behavior	Churn cases, pricing pressure, lower terminal growth	Avoid comparing to more defensible vertical platforms	Existing code may have less value if the workflow is commoditized
Infrastructure or data platform	AI demand increases usage, but cost and competition also rise	Net revenue retention, compute cost, margin stability, security, and data governance	Usage growth, gross margin, cloud commitments, customer concentration	Compare by platform depth, scale, contract quality, and cash-flow conversion	Data processing technology, platform architecture, contracts, and customer relationships
Thin AI wrapper	Product depends heavily on third-party model APIs and has limited proprietary differentiation	Churn, vendor dependence, gross margin compression, IP risk, and claims risk	Shorter forecast life, higher reinvestment, downside scenarios	Avoid an AI premium without defensibility evidence	Limited proprietary assets unless data, code, workflow, or contracts are transferable

Quick Answer: Does AI Increase or Decrease a Software Company’s Value?

AI increases the value of a software company only when it improves the economic benefits available to owners or investors after considering risk, reinvestment, and transferability. A company that uses AI to solve a costly customer problem, retain customers longer, increase expansion revenue, reduce recurring service burden, or protect a proprietary data advantage may have stronger valuation support. A company that merely adds generic AI features, depends on a third-party model vendor, or uses AI claims that cannot be substantiated may receive little or no valuation benefit.

The reason is that business valuation is evidence-based. Houlihan Lokey’s 2026 AI and software valuation framework describes AI effects as valuation considerations under both the income approach, including discounted cash flow assumptions, and the market approach, including growth, margin, discount-rate, and multiple-selection considerations (Houlihan Lokey, 2026). Oliver Wyman likewise frames agentic AI as changing how investors evaluate SaaS because software may shift from augmenting human workflows toward executing more of those workflows (Oliver Wyman, 2026). Bain’s analysis similarly warns that generative and agentic AI can enhance some SaaS offerings while replacing or replicating others (Bain & Company, 2025).

That is why a professional business appraisal should ask, “What changed?” rather than “Does the company use AI?” The appraiser should review customer behavior, financial data, product usage, churn, renewal notes, pricing tests, sales win/loss information, gross margin by product, model and cloud costs, vendor contracts, data rights, IP documentation, and management forecasts. If AI is visible in customer outcomes and financial performance, it can be modeled. If it is only visible in marketing language, it should not carry much valuation weight.

There are four possible outcomes:

1. AI is a value driver. It supports better revenue quality, margins, risk profile, or asset value.
2. AI is a risk factor. It creates substitution risk, legal uncertainty, cost pressure, vendor dependence, or customer trust issues.
3. AI is mostly noise. It is a feature or internal tool that does not materially change revenue, margins, risk, or assets.
4. AI changes the story but not yet the evidence. The company may have a credible roadmap, but the valuation still needs scenarios, sensitivity analysis, and documentation.

This is especially important in 2026 because AI adoption and corporate investment are broad enough that buyers expect diligence, but broad adoption does not prove a private company deserves a higher value. Stanford HAI describes AI’s growing social, economic, and governance influence and notes accelerating business adoption in recent years, but such macro evidence must still be translated into company-specific valuation assumptions (Stanford Institute for Human-Centered Artificial Intelligence, 2025). IBM’s AI implementation research likewise supports the idea that organizations are actively operationalizing AI, but it does not eliminate the need to examine whether a specific company’s AI strategy produces durable value (IBM, 2024).

The 2026 AI Valuation Lens: From Product Feature to Economic Engine

In earlier software valuation discussions, AI was often treated as a product feature or future roadmap item. In 2026, it is better understood as an economic lens. The appraiser, buyer, or investor should determine whether AI changes the company’s value drivers in a measurable and transferable way.

Why AI diligence is now part of software valuation

AI diligence matters because it can affect several elements of software value at once:

• The product’s ability to solve a customer problem.
• The durability of recurring revenue.
• Churn and net revenue retention.
• Pricing power and expansion revenue.
• Gross margin after inference, hosting, data, and vendor costs.
• Sales efficiency and customer acquisition cost.
• Product development requirements.
• Customer support and implementation burden.
• Data rights, privacy controls, model monitoring, and governance.
• Buyer confidence in the company’s claims and assets.

Those variables feed directly into the income approach, especially discounted cash flow. They also affect the market approach because a company with defensible AI economics may not be comparable to a thin AI wrapper, even if both use similar marketing language. In the asset approach, AI may increase the importance of developed technology, source code, proprietary datasets, model assets, patents, trade secrets, licenses, and customer contracts.

Professional valuation standards and guidance emphasize discipline, scope, assumptions, methods, and documentation. NACVA’s standards materials, AICPA and CIMA’s VS Section 100 materials, and IVSC standards resources all support the broader point that valuation work should be performed with a defined engagement, appropriate methods, and supportable analysis rather than unsupported labels (AICPA & CIMA, n.d.; International Valuation Standards Council, n.d.; NACVA, n.d.).

Agentic AI and workflow substitution risk

Agentic AI is particularly important because it raises a harder question than earlier automation tools. If AI systems can execute workflows that used to require users to navigate several software products, then some software companies may become more valuable while others become less protected. Oliver Wyman describes recent advances in agentic AI as shifting software from a tool that augments human workflows toward systems that can increasingly execute workflows end-to-end (Oliver Wyman, 2026). Bain similarly frames the issue as both an opportunity and a disruption risk for SaaS leaders (Bain & Company, 2025).

For valuation purposes, the question is not whether agentic AI is “good” or “bad.” The question is whether it deepens the company’s role in the customer workflow or bypasses it. A vertical platform that owns specialized data, integrates deeply with customer operations, and produces measurable outcomes may become more embedded. A narrow horizontal tool that automates a task now handled by generic AI agents may face a shorter competitive advantage period, more churn, and more price pressure.

Four AI positions for valuation purposes

AI position	Practical meaning	Potential valuation implication	Key diligence question
Core AI-native product	AI is central to the customer value proposition	Value may improve if outcomes, data rights, and defensibility are proven	Are customers paying, renewing, and expanding because of the AI-driven outcome?
AI-enhanced product	AI improves an existing software product	Value depends on measurable retention, upsell, margin, or productivity benefits	Are the economics better after ongoing AI costs?
AI-exposed product	AI could automate around or replace the product	Value may face lower growth, pricing pressure, or higher risk	Can the product remain embedded and differentiated?
AI-wrapper product	The company packages third-party AI with limited proprietary moat	Value may be fragile without proprietary data, workflow, distribution, or contracts	What remains if model vendors or competitors copy the feature?

This classification should be revisited during each valuation method. A company might look attractive under a revenue-growth story but weaker under a free-cash-flow lens. Another company might have limited current revenue but meaningful technology, data, or customer-contract assets. A third might have strong EBITDA today but face high reinvestment needs to remain competitive.

How AI Changes the Income Approach and Discounted Cash Flow

The income approach is often the best place to test AI claims because it forces management’s narrative into explicit financial assumptions. A discounted cash flow model values expected cash flows, their timing, and their risk. AI affects a DCF only through supportable changes in those inputs.

DCF is where AI assumptions must become numbers

A DCF should not say, “AI premium.” It should ask:

• Does AI increase revenue growth?
• Does it reduce churn or improve net revenue retention?
• Does it support higher prices or expansion revenue?
• Does it reduce support, implementation, or development costs?
• Does it increase model, inference, hosting, data, security, monitoring, or governance costs?
• Does it require more R&D or specialized talent?
• Does it change working capital, deferred revenue, cloud commitments, or capitalized software investment?
• Does it change terminal value because the competitive advantage period is longer or shorter?
• Does it change risk enough to affect scenario weighting or discount-rate judgment?

Houlihan Lokey’s framework specifically connects AI valuation considerations to income approach assumptions such as growth, margins, discount rates, and market approach selection (Houlihan Lokey, 2026). That is the right way to think about the issue. AI is not an extra line item pasted onto enterprise value. It is a set of facts and risks that should be reflected in forecast assumptions.

AI-driven revenue assumptions

AI can support stronger revenue assumptions when customers clearly receive better outcomes and are willing to pay for them. Evidence may include:

• Higher conversion rates after AI features launch.
• Higher renewal rates among customers using AI features.
• Expansion revenue tied to AI functionality.
• Reduced time-to-value in implementation.
• Customer case evidence showing measurable savings or revenue gains.
• Lower churn for cohorts that use AI features deeply.
• Contracts that separately price AI capabilities.

AI can weaken revenue assumptions when customers can replace the product with a generic model, platform-native feature, or cheaper AI-enabled alternative. Evidence may include negative renewal notes, declining win rates, high usage but low willingness to pay, customer complaints about accuracy or reliability, or competitors offering comparable AI functionality at lower prices.

The critical point is to separate adoption from monetization. Customers testing an AI feature is not the same as customers renewing because of it. Free usage is not the same as pricing power. High model usage can be a negative if it increases cost faster than revenue.

AI-driven cost and margin assumptions

AI can improve margins if automation reduces recurring support, implementation, content production, quality assurance, or development burden. It can also pressure margins if the company must pay for model inference, model hosting, cloud infrastructure, data labeling, monitoring, security reviews, testing, compliance processes, or specialized engineering talent.

That is why gross margin by product matters. A company may report attractive consolidated software margins while a new AI feature has very different economics. If the AI module drives expansion revenue but consumes expensive model calls, margin may improve less than revenue suggests. If the company capitalizes development costs, EBITDA may look stronger while cash flow remains constrained by ongoing software investment.

AI-to-DCF driver matrix

DCF input	Potential positive AI effect	Potential negative AI effect	Evidence to request
Revenue growth	New AI product revenue, higher usage, expansion sales	AI competition reduces demand or price	Bookings bridge, ARR schedule, usage data, pricing tests, win/loss analysis
Churn and retention	Better outcomes improve renewals	AI substitutes reduce stickiness	Cohort retention, renewal notes, churn reasons, customer interviews
Gross margin	Automation reduces service or support cost	Inference, hosting, model-vendor, and cloud fees rise	Gross margin by product, model cost logs, vendor invoices
EBITDA	Labor leverage improves operating profit	Recurring AI costs are excluded too aggressively	Normalization schedule, add-back support, operating expense detail
Free cash flow	Higher margin converts to cash	R&D, capitalized software, cloud commitments, and working capital absorb cash	Cash-flow statement, capex, development capitalization, vendor commitments
Terminal value	Durable workflow moat supports longer advantage period	Substitution risk shortens advantage period	Roadmap, customer interviews, competitive analysis, product usage
Risk and scenario weighting	Strong controls reduce uncertainty	IP, privacy, model, claims, or regulatory risk increases uncertainty	AI governance files, counsel memos, risk register, incident logs

DCF scenario calculation block

The following is a simplified hypothetical teaching example. It is not a valuation conclusion, not a market multiple, and not a recommendation for any particular company.

Illustrative AI valuation scenario weighting only

Base case present value of expected unlevered free cash flows:       $8,000,000
Upside AI adoption case present value:                               $10,500,000
Downside substitution / model-cost case present value:                $5,500,000

Illustrative probability weights:
Base case:       50% x $8,000,000  = $4,000,000
Upside case:     25% x $10,500,000 = $2,625,000
Downside case:   25% x $5,500,000  = $1,375,000
---------------------------------------------------------------
Indicated probability-weighted enterprise value:               $8,000,000

Valuation lesson: AI can increase, reduce, or leave value unchanged
based on evidence, risk, and scenario probabilities.

A scenario model is often more useful than a single aggressive forecast. The appraiser might model an upside case where AI increases expansion revenue and lowers support cost, a base case where the feature is useful but not transformative, and a downside case where AI competition compresses price or raises churn. The resulting value may be higher, lower, or similar to the pre-AI analysis.

Sensitivities that deserve attention

AI-affected software valuations should often test sensitivity to:

• ARR growth and conversion from pilot to paid usage.
• Gross retention and net revenue retention.
• Gross margin after model and hosting costs.
• Customer acquisition cost and sales efficiency.
• R&D and AI governance spend.
• Capitalized software development and cash reinvestment.
• Cloud commitments and vendor pricing changes.
• Customer concentration and contract length.
• Terminal growth and terminal margin.
• Discount-rate judgment or scenario-risk weighting.

Cost-of-capital inputs should be selected from appropriate market data and professional judgment. Sources such as Damodaran’s public datasets can provide market context, but a final valuation should not copy a current rate or equity-risk-premium input without confirming the valuation date, source, method, subject-company risk, and engagement purpose (Damodaran, n.d.).

EBITDA, Adjusted EBITDA, and Free Cash Flow in AI Software Valuations

EBITDA still matters in many software valuations, especially when buyers, lenders, or transaction databases use EBITDA-based metrics. But EBITDA can be misleading when AI changes cash reinvestment, cloud costs, model costs, or accounting treatment. In AI software valuations, the appraiser should bridge EBITDA to free cash flow and test each adjustment.

Why EBITDA still matters

EBITDA can help compare operating earnings before financing structure, taxes, depreciation, and amortization. It can be useful in the market approach when comparable transactions or companies are discussed using EBITDA metrics. It also helps identify whether a software company has moved beyond pure growth into recurring operating profitability.

However, EBITDA is not free cash flow. A company can report positive EBITDA while spending heavily on capitalized software, model infrastructure, security, data labeling, implementation, or cloud commitments. A company can also show weak EBITDA because it is investing in a defensible AI platform that may generate stronger future cash flow. The valuation question is not whether EBITDA is high or low in isolation. The question is whether normalized earnings and future cash flows are supportable.

AI-specific EBITDA normalization issues

AI often creates add-back disputes. Management may argue that AI pilot costs, implementation costs, consulting fees, model migration costs, or one-time integration expenses should be excluded from adjusted EBITDA. Some of those costs may be nonrecurring. Others may be necessary to operate the company in the future.

A practical rule is to ask whether the cost is required to support the forecast. If ongoing model hosting, monitoring, privacy review, security testing, prompt management, data governance, or AI product engineering is needed to sustain revenue, it generally should not be removed from normalized earnings. If a cost was a one-time migration expense that will not recur and is not needed to sustain the future product, an adjustment may be supportable with documentation.

Public-company non-GAAP materials are not direct rules for most private valuation clients, but they are a useful reminder that metric definitions and reconciliations matter. SEC materials on non-GAAP financial measures focus on public-company disclosure context, including EBITDA and adjusted EBITDA issues (U.S. Securities and Exchange Commission, n.d., 2003). For private-company business valuation, the practical lesson is to define EBITDA, adjusted EBITDA, ARR, free cash flow, and related metrics clearly and reconcile them to the company’s books.

AI EBITDA-to-free-cash-flow bridge

This simplified example shows why an AI company with attractive adjusted EBITDA may still have modest free cash flow. It is hypothetical and not a valuation conclusion.

Illustrative AI software cash-flow bridge only

Adjusted EBITDA before AI review                                      $2,000,000
Less: recurring model hosting / inference costs omitted from forecast  (350,000)
Less: recurring AI governance, monitoring, and security costs           (150,000)
Less: capitalized software development cash spend                       (600,000)
Less: maintenance capex and cloud commitments                           (100,000)
Less: normalized working-capital investment                              (75,000)
Less: normalized cash taxes                                             (225,000)
-------------------------------------------------------------------
Illustrative unlevered free cash flow                                  $500,000

Valuation lesson: AI-related EBITDA add-backs must be tested against
cash-flow reality.

The bridge also helps buyers and owners discuss value more productively. If AI improves EBITDA but consumes cash through recurring model costs and development spend, value may not increase as much as the earnings metric suggests. If AI reduces support burden and does not require significant incremental infrastructure, free cash flow may improve more directly.

Market Approach: Why “AI Multiple” Shortcuts Are Dangerous

The market approach compares a subject company to guideline public companies or transactions when the data is sufficiently relevant. AI makes this harder, not easier. The label “AI” does not establish comparability.

Market approach basics for AI software companies

In a market approach, an appraiser may consider revenue, EBITDA, ARR, gross profit, or other metrics depending on the subject company, data availability, and engagement purpose. The appraiser then evaluates comparability and makes adjustments for size, growth, margin, retention, risk, liquidity, control, customer concentration, and other factors.

AI affects this process because two software companies can both say they are AI-enabled while having very different economics. One may own proprietary workflow data and sell mission-critical automation. Another may resell a third-party model through a basic interface. A third may be a mature SaaS company using AI to reduce support cost. A fourth may be vulnerable because agentic AI can replace the task its product used to perform.

Applying a generic AI multiple to all of them is not a valuation method. It is a shortcut that can produce an unsupported conclusion. Market evidence can be useful, but it must be tested for comparability.

Market approach comparability matrix

Comparability factor	Higher-quality AI/software signal	Lower-quality AI/software signal	Valuation implication
AI role in product	AI is embedded in a critical customer workflow	AI is a minor feature or marketing label	Affects whether AI supports a stronger valuation narrative
Data moat	Proprietary, licensed, clean, transferable, and hard-to-replicate data	Generic public data or unclear data rights	Affects defensibility, risk, and asset value
Retention	AI improves renewal, expansion, or workflow adoption	AI fails to change customer behavior	Affects DCF assumptions and market quality adjustments
Margin quality	Automation benefit exceeds model and cloud costs	Inference and hosting costs compress margins	Affects EBITDA and free-cash-flow comparability
Product risk	AI deepens workflow integration	Agentic AI may bypass the product	Affects growth, terminal value, and risk
Governance and claims	AI claims are controlled and substantiated	AI-washing, privacy, or IP concerns exist	Affects diligence risk and buyer confidence
Metric quality	ARR, EBITDA, adjusted EBITDA, and free cash flow reconcile to books	Definitions change or cannot be supported	Affects reliance on market approach inputs

Public cloud and practitioner sources need caution

Venture capital commentary, SaaS benchmark reports, public cloud indices, and practitioner market updates can help appraisers understand market narratives. They should not be used as automatic valuation conclusions. Public companies differ from private companies in scale, liquidity, capital access, customer diversification, governance, disclosure controls, and risk. Transaction datasets may reflect strategic synergies, deal structure, buyer-specific considerations, earnouts, rollover equity, or incomplete information.

If a valuation uses a numeric market multiple, it should document the source, date, sample, methodology, and relevance. It should also explain why the selected metric is appropriate for the subject company. This article intentionally does not provide generic AI revenue multiples or EBITDA multiples because unsupported ranges can mislead owners and buyers.

Asset Approach: Source Code, Models, Data, and AI Intangibles

Software companies are often described as asset-light. That does not mean asset-free. AI can make the asset approach more important, especially for early-stage, distressed, IP-heavy, transaction-specific, or asset-sale situations.

Asset-light does not mean asset-free

An AI-affected software company may have many assets and liabilities that matter to value:

• Developed technology and source code.
• Capitalized software development.
• Proprietary datasets.
• Data licenses and customer data rights.
• Model assets, fine-tuning records, prompts, evaluations, and model documentation.
• Patents, patent applications, trademarks, copyrights, and trade secrets.
• Customer relationships and contracts.
• Vendor agreements and model-provider terms.
• Deferred revenue obligations.
• Working capital, debt, leases, and contingent liabilities.
• Privacy, security, IP, claims, or regulatory risks.

For a profitable going-concern software company, the income approach and market approach may receive more weight than the asset approach. But the asset approach can still provide important cross-checks. It may be especially relevant when current earnings do not capture technology value, when revenue is early but assets are valuable, when a buyer is acquiring code and data rather than the whole operating company, or when specific assets must be identified for legal, tax, accounting, or transaction purposes.

AI-specific intangible asset questions

AI raises specific questions that should be addressed before assigning significant value to intangible assets:

• Who owns the source code?
• Are contractor and employee invention assignments complete?
• Who owns or can use the datasets?
• Are customer data rights documented?
• Are data licenses transferable to a buyer?
• Can the company use the data for model training, product improvement, or resale?
• Are model weights, fine-tuned models, prompts, or embeddings owned, licensed, or controlled?
• Does the product depend on a single model vendor?
• Do vendor terms restrict use, pricing, transfer, or output?
• Are AI-generated outputs material to the company’s assets or customer deliverables?
• Are patent, copyright, trademark, or trade-secret claims supported?

The U.S. Copyright Office has issued AI-related materials addressing copyrightability, human authorship, and generative AI training issues, making IP documentation an important diligence topic (U.S. Copyright Office, n.d., 2025a, 2025b). The USPTO has also issued guidance for AI-assisted inventions, including the point that no new separate standard is created for AI-assisted inventions and that U.S. patent inventors must be natural persons (U.S. Patent and Trademark Office, 2025). These sources do not turn appraisers into IP counsel, but they show why ownership and rights questions can affect value.

Asset approach and AI intangible asset checklist

Asset or liability area	Documents to request	Valuation relevance
Source code and developed technology	Repository records, architecture documentation, development history, capitalization policy	Supports technology asset value, maintenance cost, and transferability
Proprietary datasets	Data lineage, data licenses, customer consent terms, usage rights	Affects moat, risk, and transferability
Model assets	Fine-tuning records, model cards if available, vendor terms, evaluation logs	Affects reproducibility, performance, cost, and dependency risk
Patents and inventions	Patent filings, invention assignment agreements, inventorship review	Supports IP diligence and ownership questions
Copyrightable materials	Human-authorship records, software ownership files, content records	Supports IP risk review and asset support
Trademarks and brand	Registrations, brand usage, customer recognition	Supports brand and customer relationship analysis
Customer relationships	Contracts, renewals, churn history, net revenue retention, implementation obligations	Supports income and asset-based intangible analysis
Contingent liabilities	IP disputes, privacy issues, AI claims review, regulatory memos	May reduce value or increase risk assumptions

When the asset approach receives more weight

The asset approach may receive more attention when:

• The company is early-stage and has limited revenue history.
• The company is distressed or being valued under a liquidation or asset-sale premise.
• The transaction is focused on source code, datasets, IP, or customer contracts.
• The company has weak earnings but valuable developed technology.
• There is a shareholder dispute, buy-sell matter, tax matter, litigation support need, or acquisition allocation where specific assets matter.
• AI-related liabilities or rights limitations may reduce transferability.

The asset approach does not eliminate the need for income or market analysis. Instead, it helps identify what the company actually owns, what can be transferred, what costs would be required to recreate assets, and what risks might reduce economic value.

AI Risk, Governance, IP, and Regulatory Diligence

AI risk does not automatically reduce value. Poorly understood AI risk can. A buyer or appraiser may be willing to underwrite AI upside when the company has governance, documentation, legal review, and risk controls. The same buyer may discount a company with unclear data rights, unsupported AI claims, weak model monitoring, or unresolved regulatory exposure.

Governance and risk controls

NIST’s AI Risk Management Framework is a useful official resource for structuring AI risk discussion. The AI RMF describes core functions such as govern, map, measure, and manage, and addresses characteristics of trustworthy AI such as validity, reliability, safety, security, resilience, accountability, transparency, explainability, interpretability, privacy, and fairness (National Institute of Standards and Technology, 2023). NIST’s generative AI profile further discusses risks unique to or exacerbated by generative AI and suggested risk-management actions (National Institute of Standards and Technology, 2024).

For valuation purposes, these resources should be used as diligence vocabulary, not as a universal legal mandate for every private software company. An appraiser does not need to audit a model. But the appraiser should understand enough to decide whether risk affects forecasts, margins, probability weights, discount-rate judgment, market approach comparability, or asset value.

AI claims and AI-washing

AI claims can create valuation risk when marketing gets ahead of product reality. The FTC announced Operation AI Comply in 2024, describing law-enforcement actions involving AI hype or AI technology used in deceptive and unfair ways (Federal Trade Commission, 2024). That is not a valuation standard, and it does not mean every AI claim creates liability. It does show that unsupported AI claims can create consumer-protection, reputational, and diligence concerns.

In a valuation engagement, owners should be prepared to substantiate AI claims. If the website says the product is autonomous, the appraiser or buyer may ask what tasks it performs, how it is tested, how human review works, what data it uses, what accuracy limits exist, and whether customers are told about limitations. If the claim cannot be supported, it may reduce buyer confidence even if no legal claim exists.

Copyright, training data, and AI-assisted inventions

AI may affect value through copyrightability, training-data rights, and AI-assisted invention issues. The U.S. Copyright Office’s AI materials address human authorship, copyrightability, and generative AI training topics (U.S. Copyright Office, n.d., 2025a, 2025b). The USPTO’s AI-assisted invention guidance provides official context for inventorship review (U.S. Patent and Trademark Office, 2025). Appraisers should not provide legal conclusions, but they should recognize when uncertainty may affect value.

For example, if a software company’s product depends on a dataset with unclear rights, a buyer may reduce value, require indemnities, or delay closing. If the company cannot document ownership of code created by contractors, technology transferability may be impaired. If AI-generated content is central to a product, copyrightability and usage rights may matter to forecast risk and asset value.

EU AI Act and broader regulatory risk

Regulation (EU) 2024/1689 lays down harmonized rules on artificial intelligence in the European Union (Regulation (EU) 2024/1689). A private software company that operates in or sells into relevant EU contexts may need legal review to determine whether and how the regulation affects its product, customers, obligations, costs, timing, or risk. This article does not summarize detailed requirements or deadlines because those conclusions depend on product facts and legal analysis.

The valuation effect is through economics: compliance cost, delayed revenue, restricted use cases, customer diligence, support burden, insurance cost, contract terms, or litigation risk. A company with clear documentation and counsel-reviewed compliance planning may have a stronger valuation position than a company with the same AI feature but no governance records.

AI risk matrix

Risk area	Warning sign	Potential valuation effect	Diligence documents
Unsupported AI claims	Marketing says “AI-powered” but product evidence is weak	Lower forecast confidence and buyer trust	Product demos, customer outcomes, claim substantiation
Data rights	Training or product data rights are unclear	Higher legal risk and lower transferability	Licenses, contracts, data lineage, customer consent records
Model dependency	Core product relies on one vendor API	Margin, pricing, continuity, and roadmap risk	Vendor terms, fallback plans, cost history
Privacy and security	Sensitive data is used without strong controls	Higher customer, compliance, and churn risk	Security policies, data-processing records, incident logs
Copyrightability and authorship	AI-generated assets lack human contribution records	Asset value and ownership uncertainty	Creation records, authorship records, legal review
Inventorship	Patent work involved AI without inventor review	Patent ownership or validity diligence issue	Invention assignments, patent files, counsel memos
Bias, reliability, explainability	AI outputs create customer or compliance risk	Higher support cost, churn, claims, or scenario risk	Model testing, monitoring, quality-control records
Regulatory exposure	Product may fall into regulated AI contexts	Higher cost, delay, restriction, or diligence risk	Counsel memo, compliance roadmap, customer contract review

Decision Tree: Is AI a Value Driver, Risk Factor, or Noise?

The following decision tree shows how AI should be translated into valuation logic. It is intentionally practical. If AI does not change customer outcomes, it probably should not materially change value. If it changes outcomes but not economics, the valuation should wait for evidence or use scenarios. If it changes economics and is defensible, it can support stronger assumptions. If it changes economics but creates governance or rights uncertainty, risk must be modeled.

This tree also helps owners prepare for diligence. It shows why the strongest evidence is not a product demo by itself. Strong evidence connects product capability to customer behavior, financial performance, legal rights, and risk controls.

Public-Company AI Examples: Useful Disclosures, Not Private-Company Multiples

Public-company filings can be useful because they show how large software companies discuss AI strategy, product capabilities, forward-looking risks, non-GAAP metrics, and disclosure themes. They should not be used as direct private-company valuation multiples. Public companies differ in scale, liquidity, diversification, governance, capital access, disclosure controls, customer base, and investor expectations.

What public filings can teach private software companies

The practical lesson from public filings is not “copy the multiple.” It is “define the claim, disclose the risk, and support the metric.” Private software companies preparing for a sale, financing, shareholder matter, or business appraisal can learn from how public companies describe AI-related product capabilities and risk factors. The information can help owners anticipate buyer questions, even though private companies usually do not have the same reporting obligations.

Company filing source	Verified AI-related disclosure theme	How to use the example	Limitation
Microsoft Form 10-K	Microsoft describes making digital technology and AI available broadly and responsibly	Example of broad AI positioning in a large technology company filing	Not a private-company comparable or multiple source
Adobe Form 10-K	Adobe includes AI, product-development plans, market opportunity, and trend language in forward-looking statements	Example of AI opportunity and risk language	Does not establish value for smaller creative or document software companies
Palantir Form 10-K	Palantir discusses expectations involving laws and regulations, including AI	Example of regulatory-risk disclosure	Not legal advice or a private valuation rule
Salesforce Form 10-K	Salesforce discusses Agentforce and AI-powered customer platform themes	Example of agentic product positioning	Large platform scale differs from private SaaS
ServiceNow Form 10-K	ServiceNow describes offerings involving governance, security, management of AI, and workflow automation	Example of AI-workflow positioning	Not an asset or income approach shortcut
Snowflake Form 10-K	Snowflake discusses development and use of AI and machine learning technology and customer adoption	Example of AI/data platform disclosure	Does not supply private-company valuation inputs
Datadog Form 10-K	Datadog describes artificial intelligence and machine learning capabilities in its platform	Example of AI-enabled product disclosure	Do not apply public market multiples to a private company

The cited filings are useful examples of disclosure themes, not evidence that any private software company should receive a specific valuation adjustment (Adobe Inc., 2026; Datadog, Inc., 2026; Microsoft Corporation, 2025; Palantir Technologies Inc., 2026; Salesforce, Inc., 2026; ServiceNow, Inc., 2026; Snowflake Inc., 2026).

Why public examples cannot set private-company value

A private software business valuation still requires subject-company analysis. The appraiser must evaluate the company’s revenue model, retention, contracts, customer concentration, management depth, product maturity, margin quality, cash-flow conversion, technology transferability, data rights, legal risk, and liquidity or control characteristics. Public filings can help frame diligence questions, but they cannot replace them.

Hypothetical Case Studies: How AI Can Raise, Lower, or Reframe Value

The following examples are hypothetical and simplified. They are intended to show how AI can affect valuation methods differently. They do not provide valuation conclusions, market multiples, or advice for any specific company.

Case Study 1: Vertical AI workflow platform with proprietary data

A hypothetical software company serves a specialized industry where customers must process complex documents, rules, and workflows. The company has long-term customer relationships, proprietary domain data, documented data rights, and AI functionality that reduces manual review time. Customers renew because the product is embedded in daily operations.

Under the income approach, the appraiser might support stronger revenue growth and retention assumptions if customer evidence proves adoption and willingness to pay. The forecast should still include model cost, engineering cost, quality-control cost, and governance cost. Under the market approach, the company should be compared to businesses with similar workflow depth, data advantage, retention quality, and margin structure, not to every company using AI language. Under the asset approach, the appraiser should review source code, datasets, data rights, customer contracts, developed technology, and IP documentation.

Practical conclusion: AI can support higher value when it creates durable economics and the supporting rights and assets transfer cleanly.

Case Study 2: Horizontal SaaS tool exposed to agentic AI substitution

A hypothetical productivity tool automates a narrow task within a broader business workflow. New agentic AI systems can perform that task across multiple applications, reducing the need for a standalone tool. The company still has recurring revenue, but win rates have slowed and renewal notes mention alternative AI tools.

The DCF should test downside scenarios for churn, pricing pressure, and lower terminal growth. EBITDA add-backs should not remove ongoing product investment required to reposition the company. The market approach should avoid comparing the company to defensible vertical platforms. The asset approach may show limited value if the code is easy to replicate and customer relationships are weak.

Practical conclusion: AI may reduce value when it weakens switching costs or compresses pricing.

Case Study 3: Mature SaaS company using AI to improve support margins

A hypothetical mature SaaS company adds AI-assisted support tools that reduce response time, triage tickets, and customer success burden. Renewal rates remain stable, customer satisfaction improves, and support headcount grows more slowly than revenue. The company keeps records of AI cost, model usage, and customer data protections.

The DCF may reflect improved operating margin if savings are recurring and not offset by model costs. The market approach may support a stronger quality adjustment if margin improvement is proven and sustainable. EBITDA normalization should separate one-time implementation cost from recurring AI operations cost. The risk review should still consider customer data, privacy, vendor dependency, and reliability.

Practical conclusion: AI can increase value when margin improvement is real, repeatable, and cash-flow positive.

Case Study 4: Thin AI wrapper with fast early growth

A hypothetical company builds a simple interface around a third-party large language model API. It grows quickly through marketing, but customers are on short-term contracts, proprietary data is limited, model costs are rising, and competitors can copy features quickly.

The DCF should test churn, gross margin compression, vendor pricing, and replacement risk. The market approach should avoid applying AI-native narratives without defensibility evidence. The asset approach may identify limited transferable IP if most value comes from third-party models and temporary demand. Risk diligence should focus on AI claims, data rights, vendor terms, and customer contract durability.

Practical conclusion: Growth alone may not support high value if the moat is weak and AI costs or substitution risk are high.

Case Study 5: Data infrastructure platform benefiting from AI demand

A hypothetical data infrastructure platform sees rising usage because customers need better data pipelines, governance, and observability for AI projects. Usage revenue is growing, but the company has large cloud commitments and some customer concentration.

The DCF should test usage growth, net revenue retention, gross margin, cloud commitments, and concentration risk. The market approach may consider infrastructure and data-platform comparability, but only after adjusting for scale, margin, and contract quality. The asset approach should review platform architecture, code, customer contracts, data-processing technology, and security documentation. Risk analysis should address privacy, reliability, and mission-critical dependencies.

Practical conclusion: AI demand can support value when customer usage, margins, and retention convert into durable free cash flow.

Mini case-study comparison table

Hypothetical case	Primary upside	Primary risk	Most important valuation test
Vertical AI workflow platform	Defensible workflow and data advantage	Data rights, reliability, governance	Does AI improve retention, pricing, and cash flow?
Horizontal SaaS tool	Existing customer base and product familiarity	Agentic substitution and price compression	Does the product remain embedded and differentiated?
Mature SaaS support automation	Recurring margin improvement	Recurring model and monitoring costs	Do savings convert to free cash flow?
Thin AI wrapper	Early growth and market attention	Weak moat, vendor dependence, churn	What remains if the model or feature is copied?
Data infrastructure platform	Increased AI-driven usage demand	Cloud cost, security, concentration	Does usage growth become durable cash flow?

Due Diligence Checklist for an AI-Affected Software Business Appraisal

Owners can improve valuation readiness by preparing the documents a buyer or appraiser will request. The checklist below is not exhaustive, but it covers many AI-specific issues that affect software valuation.

Financial and valuation documents

• Three to five years of financial statements and tax returns if relevant to the engagement.
• Year-to-date income statement, balance sheet, and cash-flow statement.
• General ledger, trial balance, and chart of accounts.
• Debt schedule, lease schedule, related-party balances, and contingent liabilities.
• Management forecast with explicit AI assumptions.
• EBITDA and adjusted EBITDA schedules with add-back support.
• Capitalized software and development cost schedules.
• Cloud, model, infrastructure, data, and AI vendor spend history.
• Working capital, deferred revenue, customer prepayment, and contract liability schedules.

Software and AI operating metrics

• ARR, MRR, bookings, churn, gross retention, and net revenue retention definitions.
• Revenue by product, customer segment, geography, and contract type.
• AI feature adoption, active usage, renewal effect, and upsell data.
• Gross margin by product and customer segment.
• Customer acquisition cost and sales efficiency.
• Support volume, customer success cost, and implementation time before and after AI.
• Model inference cost, vendor cost, data labeling cost, and AI monitoring cost.
• Product roadmap and roadmap execution history.

Technical and IP materials

• Source code ownership records and repository access history.
• Employee and contractor invention assignment agreements.
• Patent, trademark, copyright, and trade secret documentation.
• Data lineage, data licenses, customer data permissions, and training-data rights.
• Model architecture overview, third-party model provider terms, and fallback plans.
• Security, privacy, and data-processing documentation.
• Model evaluation, monitoring, and quality-control records.

Risk, governance, and claims support

• AI governance policy, risk register, and incident response records.
• Customer-facing AI claims, marketing materials, and substantiation files.
• Legal or counsel memos about IP, privacy, data rights, EU AI Act, sector-specific rules, or customer contract restrictions.
• Customer complaints, disputes, warranty claims, or regulatory inquiries involving AI.
• Cybersecurity and privacy review materials relevant to AI features.
• Board or management materials approving AI strategy, risk appetite, and budget.

This documentation helps the appraiser determine whether AI is a durable value driver, an ordinary operating tool, a risk factor, or a claim that needs more evidence.

Common Mistakes in AI Software Valuations

Mistake 1: Treating “AI-enabled” as a valuation conclusion

AI is a capability, feature, cost, risk, or moat depending on the facts. It is not a value conclusion. A company with weak retention, unclear data rights, and heavy model costs should not receive a stronger valuation simply because it uses AI terminology.

Mistake 2: Applying public AI multiples to a private company

Public-company scale, liquidity, disclosure, capital access, and diversification differ from private companies. A public AI-related company’s market value does not establish the value of a private company with different revenue quality, margin structure, risk, and transferability.

Mistake 3: Ignoring model and infrastructure costs

AI can increase revenue while reducing gross margin. Appraisers should review model calls, inference costs, hosting, cloud commitments, monitoring, security, data, and engineering expenses. A feature that looks attractive in demos may have weaker economics after recurring costs.

Mistake 4: Removing recurring AI costs from adjusted EBITDA

Costs required to maintain product quality, security, governance, customer commitments, and compliance readiness usually belong in ongoing economics. Removing them can overstate normalized EBITDA and value.

Mistake 5: Overlooking data rights and IP ownership

Unclear rights can affect transferability, asset value, forecast risk, and buyer confidence. Data licenses, customer contracts, contractor agreements, and invention assignments should be reviewed before significant value is assigned to AI assets.

Mistake 6: Ignoring substitution risk

Some AI products deepen workflow value. Others are vulnerable to generic AI agents or platform-native features. A valuation that assumes growth without testing substitution risk may overstate terminal value.

Mistake 7: Confusing AI adoption with willingness to pay

Users experimenting with an AI feature is not the same as customers paying for it, renewing because of it, or expanding because of it. Valuation should focus on monetization and retention evidence.

Mistake 8: Failing to document assumptions

A professional business appraisal should document facts, assumptions, uncertainty, methods, and reconciliation. AI makes documentation more important because forecasts may depend on uncertain adoption, costs, regulation, IP rights, and competitive responses.

How Simply Business Valuation Helps With AI-Affected Software Valuations

If your software company is preparing for a sale, shareholder buyout, financing discussion, litigation support need, tax planning matter, strategic review, or internal equity decision, do not rely on an AI label or generic software multiple. Simply Business Valuation can help translate AI-driven revenue, EBITDA, discounted cash flow assumptions, market approach evidence, asset approach considerations, developed technology, data rights, and company-specific risk into a supportable business valuation and professional business appraisal.

A thoughtful valuation process can help owners answer buyer and adviser questions before they become deal obstacles. It can also help management understand whether AI is improving enterprise value, consuming cash, increasing risk, or requiring clearer documentation. For founders and CFOs, that clarity is useful even before a transaction.

Simply Business Valuation’s work does not replace legal, tax, accounting, cybersecurity, IP, ERISA, securities, audit, investment banking, or lending advice unless separately agreed in writing with qualified professionals. For AI-affected software companies, those advisers may need to coordinate so that valuation assumptions, legal rights, tax positions, accounting policies, and diligence responses are consistent.

FAQ: Artificial Intelligence and Software Valuations in 2026

1. Does adding AI to a software product automatically increase valuation?

No. AI affects value only if it changes durable economics, risk, or transferable assets. A minor AI feature with no effect on retention, pricing, margin, or defensibility may have little valuation impact. AI can also reduce value if it increases costs, creates vendor dependence, or exposes the product to substitution risk.

2. How does AI affect discounted cash flow analysis?

AI can affect DCF assumptions for revenue growth, churn, pricing, gross margin, model cost, R&D, working capital, terminal value, and risk. The key is to connect AI assumptions to evidence such as customer usage, renewal behavior, price acceptance, model costs, and governance documentation.

3. Should AI software companies be valued using revenue multiples?

Revenue metrics may be relevant in some market approach analyses, especially for high-growth software companies. But revenue multiples require verified comparable data and careful adjustments for growth, retention, margins, risk, scale, liquidity, and business model. A generic AI revenue multiple is not a defensible valuation method.

4. How does AI affect EBITDA?

AI can improve EBITDA if it reduces recurring labor, support, implementation, or delivery costs. It can reduce EBITDA if model hosting, inference, cloud infrastructure, monitoring, data governance, security, and specialized engineering costs are significant. Adjusted EBITDA add-backs should be tested to determine whether costs are truly nonrecurring.

5. Why might AI reduce a software company’s value?

AI may reduce value if it weakens switching costs, allows customers to replace the product, compresses pricing, increases gross-margin pressure, creates vendor dependence, raises IP or regulatory uncertainty, or causes customer trust issues through unsupported claims.

6. What is the market approach for an AI software company?

The market approach compares the company to relevant guideline companies or transactions when data is sufficiently comparable. For AI software companies, comparability should consider product role, customer segment, revenue quality, retention, data moat, margin quality, model cost, governance, and risk. AI labels alone do not create comparability.

7. How does the asset approach apply to an AI software company?

The asset approach reviews assets and liabilities such as source code, developed technology, model assets, proprietary datasets, customer relationships, IP rights, capitalized software, working capital, deferred revenue obligations, vendor contracts, and contingent liabilities. It may receive more weight for early-stage, distressed, IP-heavy, or asset-sale situations.

8. Are model weights, prompts, or datasets valuable assets?

They can be valuable if they are owned or transferable, legally usable, technically useful, economically tied to customer value, and difficult to replicate. They may have limited value if the company lacks rights, documentation, exclusivity, or customer monetization.

9. What AI due diligence should buyers request?

Buyers should request financial statements, AI cost data, product adoption metrics, ARR and churn definitions, model and vendor terms, data rights, IP documentation, source code ownership records, governance policies, customer contracts, security documentation, and claim substantiation materials.

10. Does the NIST AI Risk Management Framework create a valuation requirement?

No. The NIST AI RMF is an official risk-management framework and useful diligence resource. It should not be described as a universal legal requirement for all private software companies. In valuation, it is useful because it helps identify governance, mapping, measurement, and management issues that may affect risk and cash flow (National Institute of Standards and Technology, 2023).

11. Does the EU AI Act affect software valuation?

It can affect valuation for companies operating in or selling into relevant EU contexts if it changes compliance cost, timing, product restrictions, customer diligence, contractual risk, or forecast assumptions. Legal conclusions about Regulation (EU) 2024/1689 should be handled by counsel.

12. How do copyright and training-data questions affect value?

They can affect transferability, asset value, risk, and buyer confidence. If ownership or usage rights are unclear, a buyer may discount value, require indemnities, delay closing, or exclude certain assets from a transaction. Copyright Office materials make these issues important diligence topics (U.S. Copyright Office, n.d., 2025a, 2025b).

13. Can public AI software companies be used as valuation comparables?

Sometimes, but with caution. Public filings can provide disclosure examples and market context, but public companies differ from private companies in scale, liquidity, governance, reporting controls, capital access, diversification, and risk. Private-company valuations require company-specific adjustments.

14. When should a software company get a professional business appraisal because of AI?

A professional business appraisal may be useful before a sale, financing, shareholder buyout, tax matter, litigation support need, equity planning decision, strategic review, or material AI-driven business-model change. It is especially helpful when AI affects forecasts, EBITDA adjustments, market comparability, data rights, source code, customer relationships, or buyer diligence.

Conclusion: AI Is a Valuation Input, Not a Valuation Shortcut

Artificial intelligence is changing software valuations in 2026, but not because every AI-related company deserves a premium. AI matters when it changes cash flows, risk, market comparability, or asset value. The same trend can support a higher value for one company and a lower value for another.

A disciplined valuation should translate AI into specific assumptions: revenue growth, churn, net revenue retention, pricing, gross margin, model cost, R&D, sales efficiency, support burden, working capital, terminal value, discount-rate judgment, market approach comparability, and asset approach support. It should test EBITDA against free cash flow. It should review source code, datasets, contracts, IP rights, governance, claims, and regulatory exposure. It should document uncertainty rather than hide it.

For owners, the best preparation is evidence. Show how AI changes customer outcomes. Show whether customers pay, renew, or expand because of it. Show the costs required to operate it. Show the rights needed to transfer it. Show the controls that reduce risk. With that evidence, AI can be a real value driver. Without it, AI may be only a story.

If your company needs a software business valuation or business appraisal that evaluates AI effects with a structured, evidence-based valuation framework, Simply Business Valuation can help you organize the financial, operating, market, and asset evidence needed for a supportable conclusion.

References

Adobe Inc. (2026). Form 10-K. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/796343/000079634326000003/adbe-20251128.htm

AICPA & CIMA. (n.d.). Statement on Standards for Valuation Services: VS Section 100. https://www.aicpa-cima.com/resources/download/statement-on-standards-for-valuation-services-vs-section-100

Bain & Company. (2025). Will agentic AI disrupt SaaS? https://www.bain.com/insights/will-agentic-ai-disrupt-saas-technology-report-2025/

Damodaran, A. (n.d.). Data for current year. NYU Stern. https://pages.stern.nyu.edu/~adamodar/New_Home_Page/datacurrent.html

Datadog, Inc. (2026). Form 10-K. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1561550/000162828026008819/ddog-20251231.htm

Federal Trade Commission. (2024). FTC announces crackdown on deceptive AI claims and schemes. https://www.ftc.gov/news-events/news/press-releases/2024/09/ftc-announces-crackdown-deceptive-ai-claims-schemes

Houlihan Lokey. (2026). Artificial intelligence and software valuations: A framework for assessing enterprise values. https://cdn.hl.com/pdf/2026/ai-and-software-valuations-march-2026.pdf

IBM. (2024). AI in Action 2024 report. https://www.ibm.com/think/reports/ai-in-action

International Valuation Standards Council. (n.d.). International valuation standards. https://ivsc.org/standards/

Microsoft Corporation. (2025). Form 10-K. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/789019/000095017025100235/msft-20250630.htm

NACVA. (n.d.). Professional standards and ethics. https://www.nacva.com/standards

National Institute of Standards and Technology. (2023). Artificial Intelligence Risk Management Framework (AI RMF 1.0). https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf

National Institute of Standards and Technology. (2024). Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile. https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf

Oliver Wyman. (2026). How AI is reshaping SaaS valuations: A guide for investors. https://www.oliverwyman.com/our-expertise/insights/2026/apr/how-agentic-ai-reshaping-saas-valuations.html

Palantir Technologies Inc. (2026). Form 10-K. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1321655/000132165526000011/pltr-20251231.htm

Regulation (EU) 2024/1689 of the European Parliament and of the Council. (2024). https://eur-lex.europa.eu/eli/reg/2024/1689/oj

Salesforce, Inc. (2026). Form 10-K. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1108524/000110852426000060/crm-20260131.htm

ServiceNow, Inc. (2026). Form 10-K. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1373715/000137371526000007/now-20251231.htm

Snowflake Inc. (2026). Form 10-K. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1640147/000164014726000008/snow-20260131.htm

Stanford Institute for Human-Centered Artificial Intelligence. (2025). Artificial Intelligence Index Report 2025. https://hai-production.s3.amazonaws.com/files/hai_ai_index_report_2025.pdf

U.S. Copyright Office. (n.d.). Copyright and artificial intelligence. https://www.copyright.gov/ai/

U.S. Copyright Office. (2025a). Copyright and artificial intelligence, Part 2: Copyrightability. https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-2-Copyrightability-Report.pdf

U.S. Copyright Office. (2025b). Copyright and artificial intelligence, Part 3: Generative AI training. https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-3-Generative-AI-Training-Report-Pre-Publication-Version.pdf

U.S. Patent and Trademark Office. (2025). Revised inventorship guidance for AI-assisted inventions. https://www.uspto.gov/subscription-center/2025/revised-inventorship-guidance-ai-assisted-inventions

U.S. Securities and Exchange Commission. (n.d.). Non-GAAP financial measures. https://www.sec.gov/rules-regulations/staff-guidance/corporation-finance-interpretations/non-gaap-financial-measures

U.S. Securities and Exchange Commission. (2003). Conditions for use of non-GAAP financial measures. https://www.sec.gov/rule-release/33-8176