Global professional services giant EY has quietly deployed an AI-powered routing system that dramatically reduces token consumption, with savings reaching as high as 60%. The move, reported by Briefs Finance, signals a growing trend among large enterprises to optimize AI-related costs as demand for large language models surges.

Behind the Hidden Router

EY's new AI router operates behind the scenes, intelligently directing queries to the most cost-effective language models without sacrificing output quality. By analyzing each request in real time, the system chooses between different AI providers or smaller, specialized models when appropriate, significantly cutting the number of tokens needed for each transaction.

This approach is part of a broader industry shift toward "model routing" or "LLM gateways," where companies use middleware to manage multiple AI services. EY's deployment is notable because it is one of the first large consulting firms to publicly acknowledge using such a system at scale, and the reported efficiency gains are substantial.

How Token Reduction Works

  • Dynamic model selection: The router sends simple queries to smaller, cheaper models and only escalates complex ones to premium LLMs.
  • Context optimization: It trims unnecessary prompt history and compresses repeated phrases.
  • Cache management: Frequently asked questions are answered from a local cache, avoiding repeated token charges.

Why Token Costs Matter for Enterprises

Token consumption is the primary billing metric for most AI APIs. Every word sent to or generated by a model counts as a token, and costs can escalate quickly when thousands of employees use AI tools daily. For a firm like EY, which has over 300,000 staff globally, even a 10% reduction in token usage translates to massive annual savings.

The 60% figure reported by Briefs Finance refers to the maximum reduction achieved in certain workloads, not an average across all use cases. Still, industry analysts note that such a level of optimization is impressive and could set a benchmark for other professional services firms.

Implications for AI Budgets

Enterprises are increasingly looking for ways to contain AI spending as adoption expands. Router technologies like EY's offer a practical solution that does not require retraining models or sacrificing accuracy. Instead, they work at the orchestration layer, making them relatively easy to integrate with existing infrastructure.

This development also puts pressure on AI providers like OpenAI, Anthropic, and Google to justify premium pricing for their flagship models. If routers can achieve near-equivalent results using cheaper alternatives, the competitive landscape may shift toward price-sensitive offerings.

AI and Blockchain Convergence

While EY's router is not blockchain-based, its deployment is relevant to the crypto and Web3 community, where token-based economies are central. The term "token consumption" in AI is distinct from cryptocurrency tokens, but the underlying concept of efficient resource usage resonates with blockchain developers who face gas fees and network congestion.

EY has been a major player in blockchain adoption, having launched several enterprise-grade solutions on Ethereum and other networks. This new AI initiative shows the firm's continued focus on emerging technologies, potentially paving the way for hybrid AI-blockchain systems that optimize both computational and transactional costs.

What This Means for Developers

For developers building decentralized applications that integrate AI, EY's approach offers a blueprint. By using routing logic to select cost-efficient models, they can reduce operational overhead and pass savings to users. Some projects are already experimenting with on-chain AI oracles, but the high cost of inference remains a barrier.

The idea of a "hidden" router also raises questions about transparency. Users interacting with AI systems may not know which model is actually answering their queries, which could have ethical and regulatory implications. However, EY's internal deployment suggests that many enterprises prioritize cost savings over disclosure.

Key Takeaways

  • EY has deployed an AI router that cuts token consumption by up to 60% in certain workloads.
  • The system dynamically selects models and optimizes prompts to reduce costs without harming output quality.
  • This move highlights the growing importance of AI cost management for large enterprises.
  • Similar routing techniques can be applied in Web3 projects to lower AI-related expenses.
  • Transparency in AI model selection may become a future concern as hidden routers become more common.