CatalystWireBeta

Update on ASIC Manufacturer AI Oracle Test

In brief · summary, not quotable

Quantum Blockchain Technologies plc has found a way to accelerate live testing of its Method C AI Oracle software by connecting it directly to an ASIC manufacturer's Mining Development Kit, bypassing full integration onto the mining rig. This approach is expected to deliver earlier preliminary results by the end of April 2026, as the company has successfully modified the rig's operating system to intercept mining jobs and pass them to the AI Oracle for more efficient Bitcoin mining. Data collection has commenced in their Milan laboratory, with neural network training and AI software model generation to follow, while a parallel project to achieve full deployment onto the mining rig continues.

Full announcement

Select text to share a quote on X · sign in to keep highlights & notes in your QBT notes

Quantum Blockchain Technologies plc (AIM: QBT), the AIM-listed investment company, is pleased to announce that it has found a way to accelerate the live testing of its Method C AI Oracle software on the mining rig of the ASIC manufacturer, the loan of which QBT announced on 6 March 2026. This work around is expected to shorten live-testing timelines and deliver earlier preliminary results of the AI Oracle which the Company now expects to receive by the end of April 2026.

The approach has been made possible by connecting the AI Oracle directly to the ASIC manufacturer’s Mining Development Kit (“MDK”) thereby avoiding the time-consuming process of fully integrating the AI Oracle onto the mining rig.

The use of the MDK circumnavigates the need to translate and optimise the AI Oracle for the ASIC control board which, whilst feasible, is a time consuming task. Instead, QBT has succeeded in modifying the CGminer-like operating system of the ASIC manufacturer’s rig to intercept all incoming mining jobs from the mining pool which are then passed to the AI Oracle with the result that only the best hashes are executed thereby enabling more efficient Bitcoin mining.

The ASIC manufacturer’s CGminer-like operating system has already been successfully tested on an Intel x86 family server in the Company’s Milan University laboratory and by its Oregon development team. This process bypasses the activities normally performed on the rig’s hashboards; instead replicating them via a cable connected directly to the server.

Data collection from the mining rig has started at QBT’s Milan laboratory. When the minimal critical mass of data has been generated, the training of the neural networks will begin, followed by the generation of the AI software model specific to the ASICs on the third-party’s hashboard.

Notwithstanding the accelerated process’ successful implementation, the objective of deployment of the AI Oracle onto the mining rig remains a key objective and the Company will be running a parallel project to also achieve this aim.

Francesco Gardin, CEO and Executive Chairman of QBT, commented

, “Connecting the AI Oracle directly to the hashboard of the MDK enables QBT’s development team to start live-testing immediately once the new AI Oracle is generated from the data collected from the mining rig. Moreover, we also believe that this solution will speed up our internal testing on the Antminer S9 and the Bitaxe Gamma mining rigs, which are both additional targets for the Company’s future customer demonstrations.

“I am, therefore, extremely pleased that this development shortcut will potentially allow QBT to provide its preliminary results earlier than expected.”

Cleaned text: letterheads, contacts and legal notices removed. View the original announcement ↗ · Company filings. Not investment advice.

Share this quote

Quote card
Post on X WhatsApp Download image

The link opens this announcement with the quote highlighted. Quotes are checked against the original text.

Add a note