United StatesFinance and insurance2025 – ongoingNasdaq
GPU embeddings for the enterprise generative AI platform
Nasdaq put GPU embeddings into production on an AI platform that 8,100 employees had tried
Problem
Nasdaq's generative AI platform faced costly embeddings and slow retrieval. Employees needed reliable answers as they began using it for daily work.
Approach
The engineering team tested self-hosted GPU embeddings during a hackathon. It then made them the primary embeddings source in production, using NVIDIA NeMo Retriever and NIM.
Results
Takeaway
The team changed the retrieval infrastructure within an existing employee platform. The disclosed user count measured people who had tried the platform, rather than regular active users.
Categories
- Industry
- Finance and insurance
- Business function
- Internal work and knowledge managementIT and R&D
- Innovation type
- Process innovation
- AI technology
- Generative AI
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Company's own claim
Sources
Source summaries
Nasdaq engineering lead Eric Reiser described the move from a hackathon experiment to production GPU embeddings. He said 8,100 employees had used the platform and thousands used it regularly.
Full details are in the original sources.