• • • 2026 PREDICTIONS • • •
MOLEX NAMES TOP 10 CONNECTIVITY AND ELECTRONICS DESIGN PREDICTIONS FOR 2026, FUELED BY FAR-REACHING IMPACT OF ARTIFICIAL INTELLIGENCES ACROSS MAJOR INDUSTRIES
Molex, electronics and connectivity innovator, predicts that artificial intelligence (AI) will continue to transform every major industry sector over the next 12 to 18 months, driving exponential demand for computational resources while causing major bottlenecks in both compute power and connectivity
A
I-driven data proliferation and processing are creating new opportunities and obstacles for design engineers across fast-growth sectors, including automotive, aerospace and defence, consumer electronics, data centres, industrial automation and MedTech. “AI models in every industry, from autonomous vehicle sensors and high-resolution medical imaging to vital defence systems and real-time control systems on factory floors, generate massive amounts of data while requiring high-speed connectivity, advanced power delivery and efficient thermal management,” said Aldo Lopez, SVP and President, Datacom and Specialty Solutions, Molex.
“In 2026, we will remain steadfast in helping remove critical bottlenecks in compute power and connectivity while working with customers, suppliers and partners worldwide to develop future-ready, AI-driven infrastructures.”
Top 10 Predictions for 2026
1. High-speed interconnects remain essential to delivering the speed and density to enable AI/Machine Learning workloads in modern hyperscale data centres
Communication between major compute elements, like GPUs and AI accelerators within a data-centre server or chassis, necessitates a mix of high-speed backplane and board-to-board solutions designed for 224Gbps PAM-4 speed, along with high-speed pluggable I/O connectors that support aggregate speeds up to 400/800Gbps while providing a path to 1.6T.
2. Energy consumption hampers data-centre scaling, propelling advancements in thermal management The heat generated by high-performance servers and systems needed to scale
38 ELECTRICAL ENGINEERING • DECEMBER/JANUARY 2026
generative AI applications while supporting the transition to 224Gbps PAM-4 has exceeded traditional solutions relying on air-cooling technology. Developments in liquid cooling, including direct-to-chip cooling, immersion cooling and passive components that enhance active cooling, will continue to gain traction and exploration over the next 12 to 18 months.
3. Demand escalates for Co-Packaged Optics to support ‘scale-up’ architectures Co-Packaged Optics (CPO) is considered essential for handling GPU-to-GPU interconnectivity in AI-driven architecture. Designed to deliver ultra-high bandwidth density directly at the chip edge, CPO enables much higher interconnect density while reducing power consumption and electrical signal loss. The focus on CPO is expected to
electricalengineeringmagazine.co.uk
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52
