Advanced Computing in the Age of AI | Friday, June 9, 2023

Keeping up with the Data Center: Will Innovation Continue? 

The enterprise and data center markets are going through what some experts call a “once in a decade transition.” Driven by the insatiable demand for ever-increasing bandwidth, these industries are rapidly moving from 40G to 100G networks while also upgrading from shorter-reach multi-mode fiber to longer-reach, single-mode fiber mesh architectures.

This is good news for the optical components companies and network equipment manufacturers because the anticipated volumes should represent a very healthy industry for many years to come, with demand likely outstripping supply at some point.

However, the question that remains unanswered is whether we can continue this rapid pace of innovation? Here we are set to deploy 100G in the second half of 2016, and the industry is already talking about 400G. The amount of money that will need to be spent in R&D to enable 400G is going to be substantial and it’s not clear the industry is set up in a way that will naturally allow that to happen. Component and module manufacturers are already investing tens of millions of dollars in manufacturing capital to support data center volumes, and when this does not get amortized, it has a negative financial impact on these companies.

Adam Carter of Oclaro

Adam Carter of Oclaro

This is further compounded by the fact that the industry wants these innovations faster than ever, placing enormous pressure on the entire distribution chain to unrealistically shorten time-to-market schedules for new data rate equipment. As hyperscale data center operators and cloud providers own both ends of the network, there has been a rapid proliferation in the number of optical interfaces and connection choices, as they can use non-standard interfaces. At 10G, there were two and at 100Gb, this has increased to eight. From a component supplier, this makes leveraging economies of scale across operations much more difficult.

To better understand this issue, let’s take a look at where the innovation in the data center is now being driven from and what needs to happen to maintain that innovation so we can continue achieving higher data rates in the future.

New Architectures Need New Technologies

Getting to higher speeds such as 100G, 200G or 400G require new, more complex network architectures in the data center. Core optics are essential elements to transition to these higher speed networks, and they are also critical for enabling the move from multi-mode networks to single-mode networks. Optical transceivers are no longer optional features in the data center. Rather, they are absolutely critical components for delivering increased faceplate density, higher bandwidth, lower power consumption and longer reach.

Just take a look at today’s mega data centers. These can house over 100,000 servers, all of which are inter-linked to function as a flexible, parallel computation engine. To realize efficient and powerful computing, interconnects between servers and switches in these data centers must operate continuously at high speeds (40G-100G); with negligible latency (nanosecond delays); low power (<35mW/G); and over long distances (10m-10km). The only way to meet these requirements is through optical fiber interconnects. When 100G is deployed, it will be driven by the availability of core optical components and modules – from the Jericho and Tomahawk ICs now available to the QSFP28 and CFP2-ACO transceivers needed to achieve the density that supports the I/O speeds on those chips.

Challenges Ahead

With 100G in sight, the industry still wants better, faster, cheaper and lower power – and they want it quicker in terms of faster time-to-market. This is going to place enormous pressure on the core optical components industry, which has been undergoing significant changes itself. Below are the key challenges and issues that the industry needs to address to keep moving forward:

  • State of the optical industry. Higher data rates and new innovations such as on-board optics are going to require enormous investments that will only be possible with a stable and healthy optical component industry that has the funds available for significant R&D investments. This will require an industry with competitive gross margins that allow it to continue funding the advancements needed for higher bandwidth, smaller form factors, higher density and lower power. It’s also important to note that the components industry is very cyclical in nature. This unevenness strains the long-term health of component companies and many times starves the required innovation needed for the next technology cycle.
  • Many different form factors and packaging formats. The rapid adoption of many different form factors and packaging formats is increasing. We need more standardization to make it easier to develop products that work through the entire supply chain. We also need to continue developing more efficient and cost-effective approaches, such as non-hermetic packaging instead of expensive metal gold boxes used in the past. These advancements will open up new doors to cost savings and added capabilities.
  • Technological innovation. It’s getting harder to deliver increasing bandwidth because we are pushing the limits of physics and you need to do this efficiently and cost effectively. Optical companies are not only going to need access to deep pockets of cash, but they are also going to require increasingly specialized expertise in core areas such as transceivers and modules so they can continue to raise the bar on their performance and functionality.
  • Re-verticalization. Today, optical companies are not only dealing with the network manufacturers like Cisco, Nokia, Juniper and Huawei, etc., but they are also now working with their customers as well. We are also seeing a re-verticalization where systems houses are acquiring small start-ups or component companies to gain access to core technology. This gives them a different business model where they give key components to suppliers or they might even go through a subcontract manufacturer where they do the design and hand it over to the contact manufacturer for manufacturing.
  • Consolidation - As bandwidth demands increase beyond 100G, it is becoming increasingly important for optical companies to have the ability to manufacture high-speed lasers and receivers capable of supporting those higher data rates. Today, there are only three to five laser fabs in the world with this capability (at >25G), and this represents an opportunity for the industry to undergo much needed consolidation.

A Path to Good Heath

Core optics are going to be critical components in the data center and enterprise for decades to come. As a result, it is going to be extremely important for optical component companies to stay ahead of the curve with smart R&D investments and business models that not only deliver the required high speed optical innovations, but long-term, profitable growth. To this end, the entire industry must evolve so the wealth is better allocated across the entire supply chain. This means changing relationships from optical component companies being solely viewed as suppliers to instead being joint partners. By working closely together to create new technologies, scalable operations and profitable business models we can raise all boats. The result will bring to market some of the most revolutionary solutions that deliver the speed, performance and value that data centers of the future need while enabling our industry to thrive and prosper.

Adam Carter is chief commercial officer of Oclaro, Inc. (Nasdaq: OCLR), a vendor of optical components, modules, and subsystems for optical transport and metro networks, enterprise networks, and data centers.