Eight sessions at the Central Asian Expo covered digital money, artificial intelligence in lending, fraud, regional interoperability, and the skills pipeline behind all of it. The second day of the inaugural Silk Road Finance & Technology Forum was programmed as Innovation Day, moving from policy discussion to implementation across keynotes on digital public infrastructure and payments reform, and panels on tokenisation, AI, cybersecurity, open finance, embedded finance and talent. The through line was sequencing, with speakers from central banks in China, Kenya, Thailand and Uzbekistan each setting out what they say has to be built, measured or regulated before digital money reaches scale.

[For more news, click here]

Two Keynotes on Timelines

Nodirbek Saydullayev, First Deputy Chairman of the Central Bank of Uzbekistan, opened the day. The first keynote came from Dr Patrick Njoroge, former Governor of the Central Bank of Kenya, who framed digital public infrastructure around verifiable digital identity, low-cost real-time interoperable payments, consent mechanisms with clear guardrails, and consent-based data exchange.

He set three tests against them, of which the second was technical soundness. He cited DhowCSD, the Kenyan platform for retail investment in government securities, which he said took five years of work before launch; published accounts describe a three-year development period. On collaboration he pointed to cryptocurrency and stablecoins, where he said conflicting US and EU frameworks risk spilling into other jurisdictions: “Coordinating regulations globally is imperative before a crisis materializes.”

Dr Veerathai Santiprabhob, former Governor of the Bank of Thailand, set out a decade of Thai payments reform. He described a 2016 baseline in which Thai banks recovered the cost of free branch and ATM channels by charging fees on the digital services meant to replace them, and electronic payments ran at roughly 60 transactions per person a year. Against that he set PromptPay, the Thai standard QR code, a national digital ID and a Payment System Act.

Dr Veerathai Santiprabhob  stated: “digital financial transformation is not a technology project. It requires ecosystem transformations that goes well beyond the technology itself. Indeed, the technology is often the easiest part to obtain.” Of his four lessons, the one aimed most directly at the host market concerned supervision. “Regulators in turn had to reimagine their own role, moving from gatekeeper to enabler.”

He also noted that not every initiative delivered on its original promise and that some were significantly delayed.

Tokenisation and Stablecoins

Moderated by Dr Peter Gassmann of PwC Strategy&, the panel opened by separating three instruments that are often conflated: privately issued reserve-backed stablecoins, tokenised assets where what matters is the underlying, and central bank digital currency as sovereign money. Mu Changchun, Director-General of the digital currency institute at the People’s Bank of China, presented monitoring data covering 36 major stablecoins across 16 public chains. On that analysis, the majority of on-chain volume consists of institutions moving funds between their own wallets, protocol-level splitting that records one piece of business as many transfers, and turnover from arbitrage and exchange activity. Transactions with a genuine payment behind them account for less than one percent. The dataset was reported in Chinese financial media in April, and a separate McKinsey and Artemis Analytics study published in January reached a comparable figure from a different methodology.

“The efficiency we’ve seen in the stablecoins and crypto assets today largely stems from the bypassing compliance requirements like customer due diligence rather than inherent technological superiority,” said Mu Changchun, Director-General of the digital currency institute at the People’s Bank of China. “Establish the rules before you pave the channels. You have to dig the pond before you reach the fish.”

Effective on-chain customer due diligence does not currently exist and that the ultimate controller of a wallet cannot be reliably identified and called for convergence through the Financial Stability Board and the Bank for International Settlements.

The Central Bank of Uzbekistan’s monetary policy director, Samigjon Inogamov, set out the domestic position. Backing requirements are settled, since tokens cannot be issued beyond available reserves. A token moving between wallets never touches a bank account, so it does not appear in the statistics the regulator uses daily, and a token used to pay a supplier is indistinguishable in transaction data from one held for a year.

Samigjon Inogamov stated: “That is why we approach to observe before we scale. And our sandbox is not only instrument to encourage innovations, but for Central Bank, it is also the instrument of measurement.”

The asks of the private sector were wallet-level flow data without names: how much moved, in which currency, and how much was held rather than spent. Context for the caution sits in the remittance figures, with inflows reaching $9.3 billion in the first half of 2026, up 13 percent, and 51.7 percent arriving as direct transfers to bank cards.

A representative of the National Agency of Prospective Projects confirmed pilots on stablecoins as a means of payment, work on tokenising bonds and shares, and a blockchain-based payment system in development, then set out the limit of current practice. “There is no state country in the world that has a fully regulated tokenization of securities market.” The unresolved problem he named was the execution of shareholder rights on a tokenised share. Liminal Custody added that institutional demand centres on collateral mobility rather than cross-border payments.

Artificial Intelligence and the Question of Control

James Boey of GFTN opened the AI panel with a short film about Aziza, a Samarkand entrepreneur refused a loan by a bank unable to establish whether the problem lay in her data, the identity check, the model or the cloud.

Davit Melikidze of UzCard argued the model layer can be international while control capabilities stay local. “We have to be able to audit, we have to be able to amend, we have to be able to change, and if something goes wrong, to recover back.”

Alexander Simonenko named identity data, decision logs, and the ability to switch providers as the things that cannot be outsourced. Sheruan Bashar of the Central Bank rejected opaque models. “I truly hope that we won’t make a loan decisions based on LLMs. So that’s not the world we live in.”

Arvind Sankaran of the Asian Development Bank noted that roughly 40 applications touch AI across a single small-business lending chain and cited a Reserve Bank of India requirement that banks submit an audit of AI use including resilience to external attack. “I would hold the lender as the primary responsible for having built those guardrails.” Yanan Wu of Surfin put the case for alternative data. “Credit score is anyone’s human right. Without credit score, you cannot start a financial journey.”

Fraud at the Speed of AI

The cybersecurity panel, moderated by Artem Saidov of KPMG Uzbekistan, turned on a specific case. Mirzabek Bobojanov of the Central Bank’s CERT unit described a fraud in which an impersonator appeared on a video call as a bank department director, and the client checked the face against the official website before transferring funds.

“The fraudster changed his face with AI and he was acting as a director of the department. And after that, the client believed to this fraudster and sent him a lot of money.” He set out a move from compliance-based to risk-based supervision, a four-tier maturity framework due next year, a requirement that all banks operate AI-based anti-fraud systems, and a threat-sharing platform launched roughly two months earlier. Nikolai Belstein of Visa argued the case for sharing. “Once one bank is under attack, meaning that the next bank will be under attack in hours, maybe in a days.” Umid Khakimov of Ipak Yuli Bank pointed to first-time users and small businesses without security staff. “We have to build the infrastructure prior to applying AI.”

Rails, Rules, and Capital

The afternoon opened on regional connectivity. Ahmet Kayhan of VEON made the telco case that connectivity precedes open finance, while Madhusudanan of M2P Fintech characterised the US as innovator-driven, Europe as regulation-driven and parts of Asia as principled and enabling.

Otabek Nasirov of the Central Asian Fintech Association described changing his position over the past year, having previously worked toward harmonised legislation across the five Central Asian states. His revised view, offered as personal rather than association policy, named four priorities: payments interoperability, mutual recognition of eKYC, API compatibility designed for regional use, and cross-border data governance.

“The objective should not be to create a single Central Asia regulator. It should be to create mutually recognized standards and interoperable infrastructure,” Otabek Nasirov of the Central Asian Fintech Association.

Khusankhodja Abidov of the Central Bank gave the regulatory sequencing. “The rules shouldn’t determine which business model wins. The rules should be created in order for all business models can compete safely… We set the course of action to regulate, but not from day one.”

His approach lets innovation proceed first where possible and escalates supervision as risk and scale increase, with the trigger set at the point where something touches personal data or critical infrastructure.

Stefan Klestil of Speedinvest said the missing element in the rails-rules-capital framing was the entrepreneur, and cited Fasset, the Dubai stablecoin banking platform that announced a $68 million round at a $1 billion valuation the previous day, led by SBI Group. He said its founders spent six or seven years obtaining licences across Kenya, Turkey, the UAE, Pakistan, Indonesia and Malaysia.

Embedded Finance and Inclusion

Dilbar Abduganieva, Chief Financial Inclusion Officer at the Central Bank, set a definitional test against the panel’s own subject. “Inclusion means that there is access, usage, and quality. The three pillars should be there.” Measured against it, her assessment was that “embedded finance is not inclusion now, because it covers only partly some of the concepts”, with transparency as the missing element.

Todd Schweitzer of Brankas accepted the point and cited Southeast Asian embedded lending at 100 percent or more interest. The regulator’s own pilot financial inclusion index, published this month, gives the gap a number: payment access scored 69 points out of 100 against a usage score of 34, with about 83 percent of adults holding a bank card but only 53 percent of cards active. Abduganieva also raised formal savings, which she said sit below both regional and global benchmarks, and asked the market to build products for it.

Rasulzhan Gulyamov of Uzum Bank described lending against marketplace transaction data for merchants without collateral or credit history. “The banking service has become very invisible.”

Dr Kanokpan Lao-Araya of the Asian Development Bank argued the collateral model is inequitable by construction.

The Talent Question

The closing session addressed the target of 5,000 fintech specialists by 2030. Asked whether the constraint is training, funding, or retention, Njoroge said training, while representatives of Ant International and the Egyptian Banking Institute said retention. Dr Abdel Aziz Nossier put the case directly: “the fintech talent is very portable. It is needed everywhere.”

Njoroge took the opposite view. “When you have talent in your own country, training people and they go to other countries, it’s not a loss.” His argument was that people return within five or ten years to senior positions, and that classroom training alone is insufficient without employers able to absorb graduates. Gulnoza Ismailova of the El-Yurt Umidi Foundation noted a change made last year under which students funded to study abroad can discharge their return obligation at large taxpayers and NGOs as well as in the public sector. The Ant International representative said the company had been asked to bring its 10x1000 training programme to Uzbekistan.

Disclosure

GFTN co-convened this forum with the Central Bank of Uzbekistan. A GFTN board member moderated the first session and a GFTN senior advisor the second, its Group Chief Executive sat on the fifth alongside the governor of the convening central bank, and the central bank stated on the fourth panel that it researched its CBDC architecture with GFTN.

Related Articles

Uzbekistan Opens Silk Road Finance & Technology Forum with a Wave of Fintech Announcements

Uzbekistan Accelerates Push to Become Central Asia’s Financial and Technology Gateway

HashKey Takes HKDAP, Hong Kong’s First Regulated Stablecoin, to the UAE