Tracking Your Preferences

I like to use my last blog of the year to focus on you, the reader. You provide valuable feedback to me and the rest of the team at Semiconductor Engineering. What do you want to see us write about? How in-depth should things be? This is always a balance between the amount of information provided and the rate at which readers tire with an article.

My focus is the channels I write for – Systems & Design and Low Power – High Performance. Perhaps to be expected this year, the readership numbers broadly align with the hot topics of the year – AI, data centers, chiplets, and multi-physics.

I do want to acknowledge my colleague Bryon Moyer, who dominated the leader board within LPHP. His more in-depth articles are clearly popular with you, and if you haven’t found them yet, you really need to check out a few of them that are listed in this blog.

AI and data centers
Everything about AI and data centers has been popular this year. Coming in third place is Lines Blurring Between Supercomputing And HPC, written by Ann Mutschler. The acceleration of performance improvements due to AI and disaggregation is driving significant changes at the leading edge of computing.

An important part of any AI model is activation functions, which Bryon Moyer wrote about in Implementing AI Activation Functions. Activation functions play a critical role in AI inference, helping to ferret out nonlinear behaviors in AI models. This makes them an integral part of any neural network, but nonlinear functions can be fussy to build in silicon.

Moyer also took the top spot in this category with New Data Center Protocols Tackle AI. This focused on developments of UALink and Ultra Ethernet. Compute nodes in AI and HPC data centers increasingly need to reach out beyond the chip or package for additional resources to process growing workloads. They may commandeer other nodes in a rack (scale-up) or employ resources in other racks (scale-out).

An emerging sub-topic associated with data centers is power and thermal. This is rapidly becoming a limiter in both what can be achieved and what can physically be built. While low-power design has not yet won over performance, it may soon become the best solution.

In third place was Moyer’s article, Will New Processor Architectures Raise Energy Efficiency? Data centers continue to heat up as new processors consume more energy than ever before. Cooling is the primary weapon against the heat these processors generate, but it wont be able to keep up forever with traditional processor architectures. New ones may be necessary.

In second place was Can Today’s Processor Architectures Be More Efficient?, another article by Moyer. For years, processors focused on performance, and that performance had little accountability to anything else. Performance still matters, but now it must be accountable to power.

The top spot went to Ed Sperling with Crisis Ahead: Power Consumption In AI Data Centers. AI data centers are consuming energy at roughly four times the rate that more electricity is being added to grids, setting the stage for fundamental shifts in where power is generated, where AI data centers are built, and much more efficient system, chip, and software architectures.

One way to deal with power issues is by considering architectural choices. Moyer considered one aspect of this with The Best DRAMs For Artificial Intelligence.  AI involves intense computing and tons of data. The computing may be performed by CPUs, GPUs, or dedicated accelerators, and while the data travels through DRAM on its way to the processor, the best DRAM type for this purpose depends on the type of system that is performing the training or inference.

In second place was a roundtable conducted by Sperling titled How AI Will Impact Chip Design And Designers. Top place was the final part of this roundtable, titled Best Options For Using AI In Chip Design.

The importance of interconnect has also risen within the broad category. Mutschler wrote about Optimizing Data Movement. Demand for new and better AI models is creating an insatiable demand for more processing power and much better data throughput, but it’s also creating a slew of new challenges for which there are not always good solutions.

Again, Moyer took the top spot with Can Cheaper Lasers Handle Short Distances? Optical technology is well established for long-haul communications, but the distances it serves are shrinking — especially in the data center.

Chiplets
Many of the advances in compute power and the ability to service the demands of AI have come from the adoption of chiplets. This is a steeply divided subject between those companies using it within a vertically integrated environment and those wishing to do it in a general marketplace.

One of the most read articles was a roundtable that I conducted at DAC this year, titled When Can I Buy A Chiplet?. This was the second annual installment of this roundtable and looked at the advances made over the past year.

Chiplets need standards, and UCIe plays an important role. I wrote Chiplets Still A Challenge With UCIe 2.0. This focuses on the needs of the few compared to those of the masses. Plug-and-play chiplets are a popular goal, but does UCIe 2.0 move us any closer to that becoming a reality? The problem is that the current drivers of the standard are not after interoperability in the way that plug-and-play requires. Perhaps I need to explore what improvements were made in UCIe 3.0?

Sperling and Mutschler put their heads together to pen Chip Architectures Becoming Much More Complex With Chiplets. The migration from monolithic SoCs to chiplet-based designs is creating a confusing array of options and tradeoffs for design teams working at the leading edge, and the number of choices is only going to increase as third-party chiplets begin pouring into the market.

The concepts of 3D-IC and chiplets have the whole industry excited, as I wrote about in 3D-IC For The Masses. It potentially marks the next stage in the evolution of the IP industry, but so far, technical difficulties and cost have curtailed its usage to just a handful of companies. Even within those, they do not appear to be seeing benefits from heterogeneous integration or reuse yet.

A similar story looked at Development Flows For Chiplets. Chiplets offer a huge leap in semiconductor functionality and productivity, just like soft IP did 40 years ago, but a lot has to come together before that becomes reality. It takes an ecosystem, which is currently very rudimentary.

In the top spot, another important part of chiplet integration was tackled by Mutschler with Signal Integrity Plays Increasingly Critical Role In Chiplet Design. Maintaining the quality and reliability of electrical signals as they travel through interconnects is proving to be much more challenging with chiplets and advanced packaging than in monolithic SoCs and PCBs.

EDA and RISC-V
While talking about development flows, verification has been making waves ever since the latest Siemens Wilson Research results were published. I wrote A Balanced Approach To Verification, which looked at some of the reasons behind this trend. It focused on the need for management to take a closer look at their verification strategies to determine if they are maximizing the potential of their tools and staff.

One of the new buzzwords for 2025 was multi-physics. I looked at this in What Exactly Is Multi-Physics? The fuzziness of the term is a reflection of just how many new and existing problems need to be addressed simultaneously in the design flow with advanced nodes and packaging.

Finally, and top spot in this category, is RISC-V’s Increasing Influence. The industry is increasingly talking about the benefits brought by the RISC-V architecture, but is it even the right starting point? While it may not be perfect, it may provide the flexibility necessary to move forward gradually.

As always, thanks for reading these and the many articles we have penned during 2025. We look forward to keeping you up to date with the emerging technologies, models, methodologies, and more in 2026. If you would like us to write about anything in particular, drop one of us an email and we may be able to add it to the publication calendar.

Happy Holidays to all. See you in 2026.