Developing
No baseline, models or measurement period were given for Mirakl's onboarding time.
In 60 seconds
- Mirakl: more than 70 million products onboarded with fine-tuned open models on NVIDIA GPUs; seller catalog onboarding now "under two hours".
- Omi claims "4x faster inference" for AI-generated product visuals built on approved 3D digital twins.
- L'Oréal's research workflows could be "up to 100x faster", a conditional ceiling, not an observed result.
- NVIDIA cites nearly $10 billion in European commitments without a breakdown.
Impact Layer
- What changed
- A marketplace software vendor is credited with automating catalog onboarding with open models.
- Why it matters
- Product data quality limits marketplace growth and agent-driven shopping, so faster onboarding matters.
- Who is affected
- Marketplace operators, sellers and commerce data teams.
- Martech stack area
- AI, Commerce
- Stack impact
- If you run a marketplace or large catalog, benchmark your own onboarding time and data-quality failure rate before accepting a vendor claim.
- Business impact
- Low Vendor roundup reported by PPC Land. Pre-deployment onboarding time, models used and the measurement period are not stated, and the marketing-relevant claims rest on vendor or customer statements.
- Related technologies
- What to watch
- Mirakl's own disclosure of methods and results.
NVIDIA published a roundup of French AI deployments on its blog on October 12, 2026, nine days before Jensen Huang’s GTC Berlin keynote, according to PPC Land.
Marketing and commerce claims (all NVIDIA-supplied)
- Mirakl: more than 70 million products onboarded using fine-tuned open models on NVIDIA GPUs; seller catalog onboarding is now “under two hours”.
- Omi: “4x faster inference” for AI-generated product visuals built on approved 3D digital twins.
- L’Oréal: research workflows with NVIDIA ALCHEMI could make discovery “up to 100x faster”, a conditional ceiling.
- NVIDIA cites nearly $10 billion in European commitments, with no breakdown by country or year.
Names without detail
Planned NVIDIA investments in Doctolib, Entalpic, Genesis AI and Pigment, with no amounts. Photoroom and Vestiaire Collective are cited for “measurable business impact” without a metric.
Not stated
Mirakl’s pre-deployment onboarding time, the measurement period and the open models used; the baseline for Omi’s 4x figure; investment sizes and closing dates.
Our read
Fast onboarding is only useful if data quality holds. Ask any vendor for error rates, not just speed.
Sources
- PPC Land: Mirakl cut seller onboarding to under 2 hours on open models, NVIDIA says (opens in a new tab)
Source for NVIDIA's blog claims and what is not stated.