EY’s Blockchain Strategy Pairs Enterprise Privacy With AI for Supply Chains

For Ernst & Young’s blockchain leaders, the enterprise case rests on two practical problems: companies need to coordinate transactions across organizations, and they can’t expose sensitive commercial details to every participant. Their approach pairs privacy tools built around zero-knowledge proofs with public blockchain infrastructure. A further step would combine that shared transaction record with artificial intelligence (AI) systems that forecast demand and manage routine supply decisions.

In an interview with Techopedia, Jeff Wong, EY’s Global Chief Innovation Officer, and Paul Brody, its Global Blockchain Leader, described the company’s work on Nightfall, a privacy system built as a Layer 2 network on Ethereum, and Starlight, a tool for turning business agreements into private smart contract logic. They also discussed where the technology is useful today, and where the combination of blockchain and AI is still more aspiration than established practice.

Why enterprise networks are shifting toward public blockchains

Private blockchains have an intuitive appeal to companies that want control over their data and transaction systems. But control can create a coordination problem: the organization operating the network may be able to charge others to use it, while its competitors have little incentive to join or help cover its costs.

Brody said companies often want to create their own networks, yet are reluctant to participate in a rival’s. That makes it hard for a private system to gather the range of partners required for transactions that cross company boundaries. EY has seen early trials work technically but stall when participants couldn’t agree on how to share the network.

Private systems can still be useful for testing. For a live business need, however, Wong said companies can usually reach their goal through a cheaper, simpler method. In his view, a private blockchain makes sense only when its specific properties justify the extra expense and its limits on participation.

Brody compares the choice to building a separate communications network for each company. Businesses now commonly run private connections over shared internet infrastructure; he argues that public blockchains can offer a similar shared foundation for transactions. The point is not that every business activity belongs on a public ledger. It’s that a network open to multiple participants is more useful when the work itself depends on those participants.

That argument has gained traction as privacy tools have improved. In the early trials Wong described, companies sometimes equated a private blockchain with privacy. Yet a private network doesn’t automatically solve the need to conceal commercial information while working with other businesses. EY’s focus has been to make those protections possible on public infrastructure, where organizations can share transaction processing without revealing the details of every deal.

Nightfall and Starlight handle different parts of a deal

Public blockchains can support tokens and smart contracts, but transactions are generally visible to network participants. For a large company, revealing what it buys, how much it pays, or where goods are going could expose sensitive business information. Brody said EY spent six or seven years developing privacy technology and, in the six months before the interview, reached a point where it could process complex transactions among multiple parties on a public blockchain with privacy, at scale and efficiency. The system supports payments and asset transfers.

The design draws on zero-knowledge proofs: mathematical methods that let a participant demonstrate that a claim is true without disclosing all the information behind it. Brody described two tools built on this foundation. Nightfall, an Ethereum Layer 2, handles private exchanges of tokens. Starlight translates the logic of a business agreement into a zero-knowledge circuit, allowing the agreement to be executed while keeping its terms private.

Consider a company buying digital switches from another business. The seller can represent the switches with tokens, while the buyer uses tokens representing money. Both can enter the transaction into Nightfall and exchange them within a shared privacy system. Other participants can see that activity is taking place, but the larger pool of transactions makes it harder to link a particular buyer and seller or infer the price of their deal.

This distinction matters because transactions aren’t all alike. A basic mixer can obscure transfers of interchangeable assets such as money, but it can’t hide the details of a unique item in the same way. Brody said non-fungible assets need a privacy system that can handle what’s being exchanged as well as the payment. Nightfall is intended to support that kind of exchange.

The exchange itself is only part of a business arrangement. Companies also need to apply conditions agreed between them. Brody gave the example of a 5% discount after a buyer spends more than $1 million. Starlight can express that kind of rule as a zero-knowledge circuit and let the parties execute it without publishing the agreement’s terms.

Put together, the tools are intended to cover both sides of a business deal: payment and goods, plus the logic governing the exchange. EY’s stated aim is to let companies conduct business-to-business agreements privately on a public network, with the transaction verified by a shared ledger.

Brody stressed that the goal is privacy, not anonymity. The system is designed so a company can disclose the details of a transaction when required while protecting other participants’ information. Zero-knowledge proofs can also provide mathematical evidence that the details shown are truthful. For enterprise users, Wong said, the complex mathematics should sit behind the service: companies should be able to make a private exchange and apply automated payments or discounts without having to manage the underlying cryptography themselves.

Operational savings may matter more than compliance

EY’s executives said the business case is often easier to make through operational efficiency than through compliance. Companies know how to negotiate a deal, Brody said, but may still rely on administrative work to carry it out. Spreadsheets, manual tracking, and repeated disputes over terms can make it difficult to apply a discount or confirm that each party has met its obligations.

A shared ledger and programmable agreement could reduce that overhead by giving multiple parties a common, verifiable record of what they agreed to do. Wong’s argument is that supply chains increasingly require companies to manage activity across the whole chain, rather than optimize only the operations inside their own business. The gains depend on participants being able to coordinate while retaining control over commercially sensitive details.

Privacy is therefore part of the operational design, rather than an optional feature. Without it, a company may avoid a shared system even if it could simplify transactions. With privacy controls, the parties can automate agreed actions and check the shared outcome without making every purchase, price, or business relationship visible to the network.

AI could add planning to blockchain’s verification

Brody sees AI and blockchain as tools with different strengths. AI can help with forecasting, planning, and communication, but its output can be wrong. Blockchain systems, by contrast, are designed to process transactions reliably and keep a tamper-resistant record of their results. Combining them could let an AI system recommend or initiate an action while a blockchain records and verifies the resulting transaction.

He pointed to AI-assisted software development as an area where imperfect output can still be useful. Generated code can be tested, debugged, and reviewed before it enters production. Some trading applications can also tolerate predictions that miss occasionally, provided the system performs well enough over time. Those settings include human checks or other ways to verify a result before it has wider consequences.

That tolerance varies by use case. Brody contrasted software development and trading with self-driving cars, where a system that is usually right may not meet the required standard. The implication for supply chain work is that teams need to decide which actions can safely be automated, what errors cost, and where a person or verification step should intervene.

His supply chain example starts at the level of an individual store. A company may make large-scale plans, but it can’t assign people to monitor every product at every location. An AI agent could track the supply of Coke cans at a Safeway store in Oakland, forecast demand, and adjust replenishment. Its forecast might occasionally send too much stock; if it’s accurate often enough, the business could still reduce inventory and improve margins.

Wong described AI-guided crypto trading as an early example of the broader pattern: AI contributes the analysis, while the transaction system provides verification. In his view, a similar pairing could eventually support supply chains, where software predicts what should happen and a shared ledger confirms what the parties actually do.

Both executives cautioned that this remains early work. AI and blockchain are still developing on their own, and their combination won’t automatically produce effective operations. Brody said there aren’t many strong real-world examples yet. The systems, data, and processes on both sides need to mature before the pairing can deliver consistently useful results.

Product identity and records across a complex chain

Supply chains involve more than moving goods between two companies. Wong described a future in which each pallet can carry an identity and associated documents, bringing together records from different systems and verification processes. A single, all-knowing platform would be difficult to build when many organizations and countries have their own systems and requirements. A decentralized network could instead help participants coordinate records across those boundaries.

Brody illustrated the documentation burden with a rule he said would take effect the following year: importers of agricultural commodities would have to show that production had not involved deforestation. Establishing that kind of claim can require extensive records from different parts of a supply chain. The example highlights why a product’s history can be as important as its location or price.

For a blockchain to help, participants would need to connect trustworthy information to the goods and preserve useful records as they move through the chain. The ledger can make shared records easier to verify, but it can’t independently establish that information entered at the source is accurate. The value depends on how identities, documents, and verification processes are linked across companies.

EY’s pitch is clearest when viewed as a progression: public infrastructure can connect participants; zero-knowledge tools can hide sensitive deal details; smart contracts can apply shared terms; and AI could help decide what action to take next. The pieces address real coordination problems, but each adds implementation work and depends on reliable information from multiple organizations. For companies considering the approach, the practical question is whether the cost of that coordination is outweighed by less manual administration, better inventory decisions, and transactions that partners can verify without exposing the full deal.