

By Raghib Hussain, President, Products and Technologies
This article was originally published in VentureBeat.
Artificial intelligence is about to face some serious growing pains.
Demand for AI services is exploding globally. Unfortunately, so is the challenge of delivering those services in an economical and sustainable manner. AI power demand is forecast to grow by 44.7% annually, a surge that will double data center power consumption to 857 terawatt hours in 20281: as a nation today, that would make data centers the sixth largest consumer of electricity, right behind Japan’s2 consumption. It’s an imbalance that threatens the “smaller, cheaper, faster” mantra that has driven every major trend in technology for the last 50 years.
It also doesn’t have to happen. Custom silicon—unique silicon optimized for specific use cases—is already demonstrating how we can continue to increase performance while cutting power even as Moore’s Law fades into history. Custom may account for 25% of AI accelerators (XPUs) by 20283 and that’s just one category of chips going custom.
The Data Infrastructure is the Computer
Jensen Huang’s vision for AI factories is apt. These coming AI data centers will churn at an unrelenting pace 24/7. And, like manufacturing facilities, their ultimate success or failure for service providers will be determined by operational excellence, the two-word phrase that rules manufacturing. Are we consuming more, or less, energy per token than our competitor? Why is mean time to failure rising? What’s the current operational equipment effectiveness (OEE)? In oil and chemicals, the end products sold to customers are indistinguishable commodities. Where they differ is in process design, leveraging distinct combinations of technologies to squeeze out marginal gains.
The same will occur in AI. Cloud operators already are engaged in differentiating their backbone facilities. Some have adopted optical switching to reduce energy and latency. Others have been more aggressive at developing their own custom CPUs. In 2010, the main difference between a million-square-foot hyperscale data center and a data center inside a regional office was size. Both were built around the same core storage devices, servers and switches. Going forward, diversity will rule, and the operators with the lowest cost, least downtime and ability to roll out new differentiating services and applications will become the favorite of businesses and consumers.
The best infrastructure, in short, will win.
The Custom Concept
And the chief way to differentiate infrastructure will be through custom infrastructure that are enabled by custom semiconductors, i.e., chips containing unique IP or features for achieving leapfrog performance for an application. It’s a spectrum ranging from AI accelerators built around distinct, singular design to a merchant chip containing additional custom IP, cores and firmware to optimize it for a particular software environment. While the focus is now primarily on higher value chips such as AI accelerators, every chip will get customized: Meta, for example, recently unveiled a custom NIC, a relatively unsung chip that connects servers to networks, to reduce the impact of downtime.
By Michael Kanellos, Head of Influencer Relations, Marvell
Computer architects have touted the performance and efficiency gains that can be achieved by replacing copper interconnects with optical technology in servers and processors for decades1.
With AI, it’s finally happening.
Marvell earlier this month announced that it will integrate co-packaged optics (CPO) technology into custom AI accelerators to improve the bandwidth, performance and efficiency of the chips powering AI training clusters and inference servers and opening the door to higher-performing scale-up servers.
The foundation of the offering is the Marvell 6.4Tbps 3D SiPho Engine announced in December 2023 and first demonstrated at OFC in March 2024. The 3D SiPho Engine effectively combines hundreds of components—drivers, transimpedance amplifiers, modulators, etc.—into a chiplet that itself becomes part of the XPU.
With CPO, XPUs will connect directly into an optical scale-up network, transmitting data further, faster, and with less energy per bit. LightCounting estimates that shipments of CPO-enabled ports in servers and other equipment will rise from a nominal number of shipments per year today to over 18 million by 20292.
Additionally, the bandwidth provided by CPO lets system architects think big. Instead of populating data centers with conventional servers containing four or eight XPUs, clouds can shift to systems sporting hundreds or even thousands of CPO-enhanced XPUs spread over multiple racks based around novel architectures—innovative meshes, torus networks—that can slash cost, latency and power. If supercomputers became clusters of standard servers in the 2000s, AI is shifting the pendulum back and turning servers into supercomputers again.
“It enables a huge diversity of parallelism schemes that were not possible with a smaller scale-up network domain,” wrote Dylan Patel of SemiAnalysis in a December article.
By Michael Kanellos, Head of Influencer Relations, Marvell
What happened in semis and accelerated infrastructure in 2024? Here is the recap:
1. Custom Controls the Future
Until relatively recently, computing performance was achieved by increasing transistor density à la Moore’s Law. In the future, it will be achieved through innovative design, and many of those innovative design ideas will come to market first—and mostly— through custom processors tailored to use cases, software environments and performance goals thanks to a convergence of unusual and unstoppable forces1 that quietly began years ago.
FB NIC on display at OFC
By Michael Kanellos, Head of Influencer Relations, Marvell
Data infrastructure needs more: more capacity, speed, efficiency, bandwidth and, ultimately, more data centers. The number of data centers owned by the top four cloud operators has grown by 73% since 20201, while total worldwide data center capacity is expected to double to 79 megawatts (MW) in the near future2.
Aquila, the industry’s first O-band coherent DSP, marks a new chapter in optical technology. O-band optics lower the power consumption and complexity of optical modules for links ranging from two to 20 kilometers. O-band modules are longer in reach than PAM4-based optical modules used inside data centers and shorter than C-band and L-band coherent modules. They provide users with an optimized solution for the growing number of data center campuses emerging to manage the expected AI data traffic.
Take a deep dive into our O-band technology with Xi Wang’s blog, O-Band Coherent, An Idea Whose Time is (Nearly) Here, originally published in March, below:
O-Band Coherent: An Idea Whose Time Is (Nearly) Here
By Xi Wang, Vice President of Product Marketing of Optical Connectivity, Marvell
Over the last 20 years, data rates for optical technology have climbed 1000x while power per bit has declined by 100x, a stunning trajectory that in many ways paved the way for the cloud, mobile Internet and streaming media.
AI represents the next inflection point in bandwidth demand. Servers powered by AI accelerators and GPUs have far greater bandwidth needs than typical cloud servers: seven high-end GPUs alone can max out a switch that ordinarily can handle 500 cloud two-processor servers. Just as important, demand for AI services, and higher-value AI services such as medical imaging or predictive maintenance, will further drive the need for more bandwidth. The AI market alone is expected to reach $407 billion by 2027.
By Michael Kanellos, Head of Influencer Relations, Marvell and Vienna Alexander, Marketing Content Intern, Marvell
Is copper dead?
Not by a long shot. Copper technology, however, will undergo a dramatic transformation over the next several years. Here’s a guide.
1. Copper is the Goldilocks Metal
Copper has been a staple ingredient for interconnects since the days of Colossus and ENIAC. It is a superior conductor, costs far less than gold or silver and offers relatively low resistance. Copper also replaced aluminum for connecting transistors inside of chips in the late 90s because its 40% lower resistance improved performance by 15%1.
Copper is also simple, reliable and hearty. Interconnects are essentially wires. By contrast, optical interconnects require a host of components such as optical DSPs, transimpedance amplifiers and lasers.
“The first rule in optical technology is ‘Whatever you can do in copper, do in copper,’” says Dr. Loi Nguyen, EVP of optical technology at Marvell.
2. But It’s Still a Metal
Nonetheless, electrical resistance exists. As bandwidth and network speeds increase, so do heat and power consumption. Additionally, increasing bandwidth reduces the reach, so doubling the data rate reduces distance by roughly 30–50% (see below).
As a result, optical technologies have replaced copper in interconnects five meters or longer in data centers and telecommunication networks.
Source: Marvell