A rewarding day in Barcelona
We’re proud to share that weeteq has been awarded Best Paper in the AI/HPC category at the OCP EMEA Summit 2026. This is a significant recognition from one of the most distinguished communities in data centre infrastructure, and it validates the depth and direction of the work we have been doing with Ultra Edge® DIVA.
“This award represents the hardwork the weeteq team have been doing behind the scenes, this highlights the monumental opportunity for weeteq in the Data Center community” – Dr Taner Dosluoglu
What DIVA delivers
Ultra Edge® DIVA is weeteq’s dual-path power delivery architecture designed for high-performance AI and GPU/xPU workloads. The architecture pairs two parallel paths – a High Efficiency Vertical Path for bulk current delivery at greater than 91% efficiency, and an Ultra Fast Parallel Regulation Path providing sub-10ns droop correction at the point of load. The Ultra Edge Controller coordinates both paths in real time, monitoring deviation from ideal transient response and compensating dynamically.
The result is that DIVA breaks the efficiency-versus-response trade-off that has constrained AI infrastructure power delivery. The efficiency path handles sustained load where it performs best. The FAST DCM path handles the transients that bulk converters cannot respond to quickly enough, with up to 100MHz droop compensation. Neither path compromises the other.
Why the parallel path matters
The 30% gain comes directly from eliminating the voltage guardbands that conventional power architectures are forced to maintain. Where a best-efficiency two-stage solution requires 130mV droop margin at 0.8V, DIVA achieves 30mV droop at 0.7V. That 100mV regulation margin translates directly into available compute performance – more power available to the processor, not burned off compensating for the limits of the power supply.
Critically, DIVA works with existing IVR and digital LDO solutions. On-chip and in-package integrated voltage regulators can be combined with the parallel path delivery system from the outset. DIVA acts as a credible upgrade path to 800V DC environments without redesigning the GPU tray or the software stack above it.
Validation that means something
This award is external confirmation that the Hyperscaler community widely recognise the problems that DIVA looks to address.
The work behind DIVA spans power electronics, silicon architecture, control system design, and systems integration. Every person on the weeteq team has contributed to getting here, and this recognition belongs to all of them. It is a direct acknowledgement that the depth of commitment and the conviction that there is a better way to deliver power to AI compute – all of it has produced something that stands up in the room where it matters.
There is more to do. But today we take a moment to recognise what the team has achieved. Watch this space.
If you are interested in reading the paper or finding out more about DIVA please fill in the form below and the team will get back to you.
