Wing Venture Capital

Develops software for efficient AI inference across diverse hardware types

Wafer runs AI inference infrastructure that adapts a model-serving stack to each customer's workload. It benchmarks and tunes models, engines, kernels and available hardware together, then keeps optimizing live serving for latency, reliability and cost as traffic changes.

Synthefy builds foundation models that generate predictions from structured data without requiring task-specific training. The models are designed for datasets such as business tables and time series, where teams need forecasts from organized numerical and categorical information.

Pioneers ASIC over Dielectric interconnect for high-bandwidth, energy-efficient AI infrastructure connectivity

Uses interpretability to understand, learn from, and design AI systems