AI Oracle Test Update

Quantum Blockchain Technologies plc (AIM: QBT), the AIM-listed investment company focused principally on a research and development programme within blockchain technology, provides a progress update on the development of its Method C AI Oracle software for the ASIC manufacturer's mining rig and Mining Development Kit (“MDK”).

Further to the Company's announcement of 8 June 2026, QBT is pleased to confirm that the first versions of the AI Oracle have now been successfully generated using the structured operational dataset collected from the ASIC manufacturer's mining rig.

As previously announced, the generation of this dataset represented a key milestone, enabling the Company's AI learning models to be trained using data produced directly by the ASIC manufacturer's mining platform. Following the successful generation of the initial AI Oracles, the Company's research team at the University of Milan has been focusing its efforts on refining the learning process to account for the unique technical characteristics of the new ASIC architecture.

Although the data generated by the ASIC manufacturer's mining rig are of the quality required for AI training, they differ significantly from the operational data previously generated by the Company using the Bitaxe Gamma platform based on the Bitmain BM1370 ASIC, both in structure and in their statistical characteristics. As a consequence, a substantial part of the recent R&D effort has been dedicated to retargeting the learning objectives of the AI models, enabling the AI Oracle to exploit the distinctive computational behaviour of the ASIC manufacturer's SHA-256 architecture.

The Company is pleased to report that this work has resulted in the successful generation of the first AI Oracles specifically trained on the ASIC manufacturer's platform. The R&D team is now focused on further improving its predictive performance before commencing live testing.

As part of this optimisation programme, the Company has also developed and tested a more compact implementation of the AI Oracle alongside the original version. Both implementations are currently being evaluated to determine the most effective balance between predictive capability, computational efficiency and deployment characteristics.

Once the performance of the AI Oracle has reached a threshold determined to be satisfactory by the Company's R&D team, it will be deployed onto the ASIC manufacturer's MDK platform for live testing under operating conditions connected to a live mining pool.

In parallel, the Company and the ASIC manufacturer's engineering team are scheduling a technical review meeting to assess the progress achieved to date and to agree the next stages of the development programme. During this meeting, an updated project timeline will be presented, reflecting the completion of the learning model retargeting phase and the remaining activities leading to live testing.

Francesco Gardin, CEO and Executive Chairman of QBT, commented:

"The successful generation of the first AI Oracles trained directly on data from the ASIC manufacturer's mining rig represents another important milestone in the development of Method C.

While we anticipated that adapting our AI learning framework to a new ASIC architecture would require additional work, the extent of the differences between the operational characteristics of the manufacturer's ASIC and those of the BM1370 platform has meant that a significant amount of effort was devoted to redefining the learning objectives of our AI models. This work has now been successfully completed, allowing us to generate the first AI Oracles specifically optimised for this architecture.

We are now concentrating on further improving the AI Oracle's performance and have also evaluated a more compact implementation alongside the original version. Once the performance reaches our target levels, we will commence live testing on the manufacturer's Mining Development Kit connected to a commercial mining pool.

We continue to enjoy a close and constructive collaboration with the ASIC manufacturer's engineering team and we look forward to presenting the next phase of the project following our forthcoming technical review meeting."