The bet is that AI adoption pushes enterprises and model builders from one-off labeling to continuous, production-grade dataset + evaluation operations, and Innodata becomes a default execution partner for the hard parts (edge cases, safety,
multimodal, regulated workflows). If it meaningfully reduces customer concentration and shifts mix toward repeatable, auditable “
model readiness” programs (retainers + metered throughput + compliance reporting), it can keep a
premium valuation versus generic IT/BPO peers despite automation pressure. That combination supports a 2–5x outcome over five years, with most of the upside coming from durable program expansion and mix/quality
re-rating rather than pure
multiple expansion.