The market for network functions virtualization (NFV) is heating up due to growing demand for IT technologies to virtualize entire network node functions. By chaining these functions together into connected building blocks, network managers and IT admins can easily roll out new communication services for their end users, creating a powerful competitive advantage.

It’s clear that a huge market opportunity is emerging for NFV enablement. NFV revenue is projected to top $37 billion by 2021, according to a recent estimate by IHS Markit. Along the way, enormous sums will be spent on the buildup of 5G mobile networks, creating new market winners and losers.

When the dust finally settles, a fresh category of competitors and startups is likely to emerge, and some of them could become the next great brands of the technology industry. One looming question is which of the current established vendors will continue to succeed? This question is especially relevant given the large-scale layoffs by some big network equipment providers, including industry giant Ericsson.

We expect several clear trends to serve as the framework for this market expansion over the coming years. They include a transition from NFV development to a deployment phase and a shift in market control from network vendors toward communications service providers (CSPs). In addition, several technology gaps still need to be overcome, and a greater focus must be put on the underlying hardware infrastructure that will serve as the foundation for NFV.

The NFV market is exiting a five-year development phase and transitioning into a five-year buildup of the full production phase. In 2012, ETSI began to formalize industry standards around the concept of near-field communications. It took roughly five years for organizations to develop these new standards and to codify the related processes.

In turn, OpenStack and other virtualization layers are now being considered as production-ready by many large-scale CSPs. Today, we have reached an inflection point whereby the industry has developed sufficient capabilities and standards for CSPs to commit to implementing NFV strategies. Now the harder question becomes, what’s the best way to get there?

The next five years will involve a rollout leading up to the widespread deployment of 5G technology. This transition will begin with the adoption of tactical point solutions that provide proofs-of-concept for network operators. It will culminate in the comprehensive full-scale replacement of legacy network technologies by NFV-enabled infrastructures.

For many years, the biggest networking vendors have wielded great control over what services the CSPs could deploy and how they could be deployed. We are now reaching a transition era in which CSP customers are taking greater control of their own destinies.

One gating factor for widespread adoption of NFV so far has involved a lack of commitment on the part of networking vendors to enable comprehensive interoperability. Many have favored building their own internal network stacks to the exclusion of integrating with technologies from outside vendors.

Until recently, this approach has been adequate for basic proofs-of-concept. But now that CSPs are moving into full production mode with NFV, they increasingly realize the importance of interoperability in which the network vendors work together. It is no longer acceptable to be thwarted by complex integrations of virtualized network functions from disparate competing vendors.

In addition, recent contributions of CSP technologies to standards organizations such as the Linux Foundation and ETSI are a sign that the largest CSPs will force their suppliers to provide smoother NFV interoperability. Networking vendors that do not get onboard with this integrated approach will ultimately find it difficult to succeed as their base businesses will start to erode and decline.

Despite so much recent progress in NFV, some significant technology problems remain to be solved. First among these concerns is the readiness of virtualized infrastructure managers to control the virtualized business layer for carrier-class operations.

Up to this point, most of the industry has been focused on making OpenStack more robust for the operational needs of carriers. In this climate, VMware has an excellent opportunity to capture greater mindshare due to the company’s well-established reputation for reliability among enterprise customers.

Another remaining gap exists in NFV automation. NFV networks can become extremely convoluted very quickly, eliminating the capacity for human oversight to manage them in a consistent, error-free manner. New technologies are needed to deliver intelligent, machine learning-based automation that can tightly control network processes.

As mentioned above, interoperability remains a significant area of focus, not just between the networking components from various vendors, but also between the layers for management components, orchestration technologies, and good old-fashioned hardware.

That brings us to the market’s insufficient focus on hardware elements. Many open source-oriented projects have focused only on software so far. The reality is that for NFV solutions to be considered carrier-grade, they will have to operate at a six-nines level of reliability. In other words, the underlying hardware must be made airtight.

All these reliability concerns cannot be left for OpenStack or the software layer to solve. There must also be an additional emphasis on correctly configuring the underlying servers and sizing them correctly to ensure the reliability of the hardware and the accuracy of the data.

In addition, these networking environments will remain hybrid for a long time because there is no single installed physical plant of existing network equipment. Therefore, for NFV projects to succeed, integration must be incorporated across the hardware aspects as part of the overall design picture.

Due to the lack of changes in the CSP infrastructure marketplace for years, there has been insufficient venture capital investment in new companies that can drive market innovations. But with the pending rollout of 5G networks and the expansion of the Internet of Things, we expect a pack of breakthrough startups to emerge in the NFV space.

These “wave two” companies will rethink the standard approach to deliver NFV solutions that were born out of the standards efforts of the past five years. These newer companies will push the industry forward by automating complexity and implementing machine-learning solutions to bridge the remaining technology gaps.