6 November 2026International edition
Vol. I · No.
6 November 2026
AI IN FASHION
DAILY
The daily briefing on AI in the fashion business
Where fashion meets artificial intelligence.
Strategy, Data & Regulation · Stories

AI vendor M&A risk assessment: 10 questions before signing a 3-year contract

Fashion design and production software saw 12 major acquisitions between 2024 and 2025. This checklist helps buyers assess acquisition likelihood, data portability, and exit clauses.

a man standing in a room next to a table
Photo: Jud Mackrill / Unsplash

KEY TAKEAWAYS Summary by the editors

  1. Fashion design and production software saw 12 major acquisitions between 2024 and 2025, driven by PLM providers acquiring niche AI and 3D startups.
  2. 30 per cent of enterprise AI contracts lack data portability clauses, and 78 per cent omit model deprecation notice provisions.
  3. Change of control clauses typically grant buyers 90 to 180 days to terminate at original pricing after a vendor is acquired.
  4. Standard AI vendor contracts provide zero to 90 days to retrieve data at termination, with some requiring requests within 30 days.

Why does vendor stability matter in a consolidating market?

Fashion design and production software saw 12 major acquisitions between 2024 and 2025, largely driven by PLM providers acquiring niche 3D and AI startups to broaden capabilities and maintain competitive positioning. Tech-driven deals such as Browzwear-Lalaland.ai and Perfect Corp.-Wannaby underscored growing demand for platforms that boost engagement, reduce development costs and accelerate product cycles. Zozo acquired fashion platform Lyst for $154 million in May 2025, whilst Zalando acquired 3D tech startup DeepAR in April 2025.

More than 35 per cent of fashion executives already use generative AI in daily operations and rank AI as the industry's biggest opportunity, according to McKinsey and Business of Fashion's State of Fashion 2026 report. Yet the risks that matter most for enterprise software relationships often emerge years into a contract: financial deterioration, acquisition by a private equity firm that dramatically changes commercial behaviour, or growing concentration risk as a single vendor becomes embedded in more and more critical processes.

What contract clauses protect you if your vendor is acquired?

Critical vendor relationships should be evaluated for contract terms, concentration risks, and transition implications, as key vendor contracts may contain change-of-control provisions affecting the transaction. On change of control, the customer gains a defined window (commonly 90 to 180 days) to terminate the contract at the original pricing through the original term, with the data portability and audit log export obligations intact.

Before signing, check whether the agreement defines what constitutes a change of control. Does it cover acquisition by a competitor? Sale to private equity? Merger with another vendor whose tools you have rejected? Negotiate advance notice requirements, typically 30 to 60 days, and the right to terminate without penalty if the acquiring entity has not been approved in writing.

How long do you have to export your data at contract end?

Standard AI contracts give you between zero and 90 days to get your data out. The weakest exit paper is Anthropic's Commercial Terms (effective 17 June 2025): data return only through the incorporated DPA and only if you ask within 30 days. Thirty per cent of enterprise AI contracts lack data portability clauses, and 78 per cent omit model deprecation notice provisions.

A detailed exit management clause is essential: the vendor must guarantee a swift, secure extraction of all your historical data, custom configurations, and logs in a readable, standardised format. Portability often means only raw input files or account records, not prompts, workflows, evaluations, logs, policies, or tuning artefacts. Specify which assets you need: training data, fine-tuned model weights, vector embeddings, API logs, and configuration files. Document format (CSV, JSON, Parquet) and delivery method (S3-compatible API, SFTP) must be named in the contract, not left to vendor discretion.

What happens to custom models and prompt libraries if you switch vendors?

Standard export clauses cover customer data; they rarely cover the operational artefacts that represent years of model tuning. Thirty-five per cent of enterprise AI contracts lack clarity on intellectual property ownership of fine-tuned models. If your team has built a library of product-description prompts, fine-tuned a sizing recommendation model, or created synthetic training data for quality control, who owns those assets?

Contracts should state explicitly that all derivative works, including fine-tuned models, retrieval-augmented generation indexes, and prompt templates built from your data, remain your property. The vendor must return or delete these on termination and provide written certification of deletion, including from backup systems. An explicit prohibition is essential: the vendor shall not use customer data, including prompts, inputs, and outputs, to train, retrain, fine-tune, or otherwise improve any of its commercial or publicly available models without prior express written consent.

How much notice will you receive if the vendor retires the model you rely on?

Anthropic's dated history shows the pattern: Claude Sonnet 4.5 was deprecated on 30 September 2026 and retires 30 November 2026. Model deprecation can break production workflows with as little as 60 days' notice. If your merchandising team has built seasonal forecasting logic around a specific vision model, or your design team relies on a particular image generator, sudden retirement forces costly rework.

Negotiate minimum notice periods for model retirement (120 to 180 days is reasonable), continued availability of the deprecated model during transition, and price protection if you are forced to migrate to a more expensive successor. The clause should specify what constitutes a material change: not only full deprecation but also changes to accuracy thresholds, API response times, or input format requirements.

What financial and operational diligence should you conduct before signing?

Vendor selection and contracting should address training data provenance, permissible commercial uses, security requirements, and indemnities for intellectual property claims. Ask for evidence of financial stability: annual revenue, funding rounds, burn rate, and runway. For venture-backed vendors, review cap tables and investor composition. If the vendor is majority-owned by a single investor with portfolio companies that compete with your business, there is potential conflict.

Prioritise contracts with the highest annual spend, known change-of-control clauses, and agreements approaching renewal dates, as well as any contract with termination penalties exceeding £50,000, as these represent the highest near-term financial risk if triggered by acquisition.

Risk categoryContract clause to reviewMinimum acceptable standard
Change of controlTermination rights on acquisition90-day termination window at original pricing
Data portabilityExport format and timelineMachine-readable format, 30-day return window
Model deprecationAdvance notice of model changes120-day notice; continued access during transition
Training data useProhibition on training with customer dataExplicit opt-out; written consent required
IP ownershipOwnership of fine-tuned modelsCustomer owns all derivatives of customer data
Exit supportTransition assistance post-termination90-day assistance at current hourly rates
Financial disclosureAnnual revenue and funding statusAudited financials or investor-backed proof of runway
Liability capMaximum vendor liabilityMinimum 12 months' fees or actual damages
Service levelAccuracy floor and uptime guaranteeDefined accuracy threshold triggering service credits
Regulatory complianceGDPR, AI Act, audit rightsThird-party audit rights; breach notification within 72 hours

What does the EU Data Act mean for AI vendor switching?

The EU Data Act's switching rules, summarised by DLA Piper (7 July 2026), cap the notice period at two months, set a 30-day transition period that can stretch to seven months only if the provider justifies it within 14 working days, give at least 30 days to retrieve exportable data, and ban switching charges, including egress, from 12 January 2027. These provisions apply to cloud and data processing services, including many AI platforms offered as SaaS.

If your vendor operates in or serves customers in the EU, verify that the contract aligns with Data Act portability obligations. The vendor may not charge you to export your data, may not impose technical barriers to switching, and must provide data in a structured, commonly used, and machine-readable format. For fashion businesses subject to environmental and human rights due diligence regulation, check whether the AI vendor's compliance documentation is portable and whether the vendor will cooperate with audits required under the Corporate Sustainability Due Diligence Directive.

How do you test exit provisions before you need them?

Conduct an exit drill for one representative workflow. Before go-live, request a test export of sample data in the contractually agreed format. Confirm that the export includes all specified artefacts: not only product images and descriptions but also metadata, training logs, model configurations, and API call history. Load the export into a staging environment and verify that it is usable without vendor-proprietary tools.

Document the time required, the technical steps involved, and any gaps between contract language and actual deliverables. If the contract promises CSV export but the vendor delivers a nested JSON structure requiring custom parsing, renegotiate the format before signing. If the vendor cannot demonstrate a working export process during proof of concept, treat that as a red flag. In late 2025 a mid-sized European dealer contracted a cloud AI platform with a data portability clause requiring S3-compatible exports and 90-day staging; when the vendor shut down operations unexpectedly in early 2026, the dealer executed the clause and retrieved all inventory metadata and high-resolution images within 21 days using the provided export scripts.

Who inside your organisation should review the AI vendor contract?

AI vendor contracts require cross-functional review. Legal counsel should assess liability, indemnity, and regulatory compliance. IT and security teams must evaluate data residency, encryption standards, and incident response obligations. Finance should model the total cost of ownership, including potential exit costs, re-implementation expenses, and switching penalties. Procurement should benchmark terms against industry standards and negotiate volume commitments and price escalation caps.

The business owner, whether head of merchandising, design, or e-commerce, must confirm that the solution aligns with operational requirements and that the team has the skills to manage a vendor transition if needed. If the vendor is core to customer-facing operations, involve your data protection officer to review training data restrictions, GDPR compliance, and customer consent workflows. If the tool generates content, involve brand and legal teams to assess intellectual property risk and approve acceptable use policies.

What questions should you ask in every vendor reference call?

Ask existing customers how the vendor handled contract renewals: did pricing increase materially, and were legacy features deprecated? Ask whether the vendor's roadmap commitments were met and how responsive support was during incidents. Ask directly whether any customer has successfully executed an exit, how long it took, and whether the vendor cooperated or created obstacles.

If the vendor has been acquired, ask customers who signed before the acquisition whether service quality, pricing, or strategic direction changed. If the vendor is venture-backed, ask whether customers have visibility into funding rounds and financial health. If the reference customer is in a different industry, ask how the vendor handles sector-specific requirements and whether the platform supports multi-tenancy with appropriate data segregation. Document the answers and compare them to the vendor's written representations.

  • Review change-of-control termination rights and ensure you can exit at original pricing within 90 to 180 days if the vendor is acquired.
  • Specify data export format, timeline, and scope in the contract; test the export process during proof of concept.
  • Negotiate model deprecation notice periods of 120 to 180 days and confirm ownership of fine-tuned models and prompts.
  • Conduct financial due diligence on venture-backed vendors: review funding, burn rate, and investor concentration risk.
  • Verify EU Data Act compliance for vendors serving European customers; switching fees are banned from January 2027.
  • Involve legal, IT, finance, procurement, and business owners in contract review to assess liability, security, cost, and operational fit.
  • Ask reference customers whether anyone has successfully exited the platform and how the vendor handled the transition.
  • Document service level agreements with defined accuracy floors and uptime guarantees that trigger credits, not just apologies.
  • Prohibit training on your data without written consent and require deletion certification at contract end.
  • Assess concentration risk: if the vendor is embedding into multiple critical workflows, negotiation leverage declines over time.

Frequently asked questions

How long does a typical AI vendor data export take?

Standard AI contracts provide zero to 90 days to retrieve data at termination. Test export processes during proof of concept to confirm that the vendor can deliver all contractually promised artefacts, including training data, model configurations, and logs, in machine-readable formats within the agreed timeline.

What is a change-of-control clause in an AI vendor contract?

A change-of-control clause grants the customer the right to terminate the contract, typically within 90 to 180 days at original pricing, if the vendor is acquired, merges, or undergoes ownership changes. The clause should define what events trigger the right, require advance notice, and preserve data portability obligations.

Do I own the fine-tuned models built from my product data?

Thirty-five per cent of enterprise AI contracts lack clarity on intellectual property ownership of fine-tuned models. Contracts should state explicitly that all derivative works, including fine-tuned models, retrieval indexes, and prompt libraries built from your data, remain your property and must be returned or deleted with written certification at termination.

Stories