Tom Bartlett
President and Chief Executive Officer at American Tower
Thanks, Adam. Good morning everyone. My focus for today's call will be on technology trends and network investments to drive demand for our leading tower and data center platforms, as well as the developments we're seeing at the edge. Well, my comments will largely be focused on the 5G evolution and the progress we're seeing in the United States. We believe similar trends will prevail across our international footprint as they have historically.
Beginning with our macro tower business, the fundamental factor that drives demand for our global power portfolio growth and mobile data consumption continues unabated. This is true both in the United States and across the globe where mobile network data traffic has almost doubled over the last two years alone to a staggering 126 exabytes per month. Looking out over the next five years, forecasted growth and data traffic per device remains compelling as more spectrum for 5G networks will be deployed at scale.
Average monthly data usage per smartphone across our key developed markets, like the US, Germany and Spain is expected to grow at a healthy compounded annual rate of 18% between 2023 and 2028. And I would note that these estimates have been somewhat conservative historically. So let's spend a moment on where we are in the 5G investment cycle in the US and where we believe we're going over the next several years. Just as we saw with the 3G and 4G rollouts, we expect the 5G investment cycle to play out in three phases that represent discrete business cases for the carriers.
And these three phases will drive to peak periods of spin that are bridged by a temporary phase of more moderate activity. The first phase is coverage driven and aimed at upgrading existing infrastructure with new spectrum bands and radio technology is competition to provide broad nationwide coverage with the new GE ramps up. At the same time, carriers are looking to realize the efficiency benefits of their investments in new software hardware and upgraded user devices.
Initial equipment upgrades and new spectrum deployment quickly deliver reduced cost per gigabyte, resulting in the ability to maintain margin profiles. Absent this migration, any incremental investments in the prior generation would be expected to result in significantly diminished returns as the additional densification required to sustain increasing network traffic on existing spectrum bands would be cost prohibitive.
As the cadence of initial coverage investments begin to moderate from record spend of over $40 billion in 2022, the first peak of the 5G cycle, we retain a high degree of conviction that there's a long tail of network investment to come. This belief is predicated on several factors, including our experience with past investment cycles industry forecasts for growth and mobile data consumption that apply necessity for significant incremental coverage and capacity and the visibility into network needs we get through our contract structures.
Today phase one of the 5G rollout is winding down and we're heading into a second phase. We expect phase two to be characterized by carriers beginning to harvest the network efficiency benefits of their initial investments, while moderating spend from the record levels of 2022 to roughly $35 billion in 2023, which is R5 billion in excess of 4g averages representing the second highest level of annual spend on record.
In this next phase, we will begin to see a seeping in of 5G technology across the wireless and enterprise landscape. For example, 5G smartphone penetration has now surpassed the 50% mark in North America, which will ultimately allow for the majority of network traffic to shift over to 5G networks, which we'd expect to occur in the 2025 timeframe. We're also looking forward to the emergence of more ubiquitous accessibility of standalone 5G core networks, which will unlock improved 5G network quality, higher speeds and lower latency and provide a platform for the development of innovative services and consumer applications.
Finally, we anticipate that the arrival of end-to-end 5G capabilities will facilitate additional monetization opportunities at the enterprise level, through use cases like private networks, network slicing, and other IoT services that are beginning to emerge today. Ultimately, these dynamics will culminate in a third capacity focused phase and significant densification of 5G networks.
We continue to believe that 5G will advance and enable the next generation of mass market consumer use cases, particularly once three GPP released 17 and 18 are in the market coupled with 5G cores that provide the true benefits of the end-to-end technology upgrade at scale. That said, meeting industry forecasts for growth and mobile data consumption that will drive the need for substantial network capacity investments seems highly achievable when taking into account the technology we have at our fingertips today.
In fact, industry estimates already showed that 5G subscribers are consuming two times to three times more mobile data than the average 4G subscriber. So let's take the case of mobile video consumption, which has consistently shown to be a dominant use case across subscriber usage types. As you can see on slide sixs, today the average smartphone subscriber in North America utilizes roughly 21 gigabytes of mobile data per month, and this is expected to grow to about 48 gigabytes by 2028.
Of the 21 gigabytes consumed today, the majority are approximately 19 gigabytes are attributed to video streaming, which corresponds to a little over an hour of daily video usage, and 360 and 480 pixel videos currently make up around half of that time. So by simply assuming a modest level of incremental usage towards higher resolution streaming such as eight minutes and 4K Ultra HD, we'd see video consumption alone drive usage to the forecasted 48 gigabytes per month, or approximately 2.3 times the current rate.
Furthermore, the data already shows that 5G is driving increased usage of higher resolution video formats. A recent report from Ericsson found that since 2021, 5G users report a nearly 50% increase in time spent on enhanced video formats. For example, among that user base, usage of new video formats like 360 degree videos, and multi view streaming have increased by an average of 10 minutes and 15 minutes per day respectively, while time spent streaming videos and standard resolution has decreased by 23 minutes over the same period.
In short, we've already seen 5G adoption linked with a shift in behavior toward using more data intensive applications. A trend we firmly believe will continue going forward. And while we remain confident that new low latency high bandwidth consumer applications will be born as 5G standalone networks are deployed at scale, we see a highly tangible case for densification requirements from where we stand today.
With that, I'll briefly provide an update on CoreSite in our data center segment before shifting to the progress we're making at the Edge. The case remains the demand in CoreSite interconnection centric business is exceeding our initial expectations. Our team's delivered record signed new business in 2022, a record we are targeting to exceed in 2023. We've also seen consistent elevated growth in interconnection revenue mark to market pricing increases that exceed our historical averages, low churn and ongoing performance that we believe positions us to deliver compelling results in the segment for many years to come.
And much likely seeing in our tower business, the secular trends that underpin the business model, like the migration of workloads from on prem to hybrid, multi cloud environments and the emergence of AI use cases that will drive more demand in our ecosystem continue on a path toward long-term acceleration. For example, findings from our recent 2023 state of the datacenter report showed that 94% of IT leaders noted that native direct connection between Colocation Data Centers major cloud providers, which CoreSite provides is essential for improved performance, enhance security, cost savings, and hybrid cloud connectivity.
Further, 92% of IT leaders are considering moving critical workloads from public cloud to colocation to accelerate revenue growth and support the increasing need for AI and machine learning applications. In this context, we continue to upgrade our offerings and capabilities within the business to support emerging use cases. For example, earlier this month, we launched new capabilities on our OCX, our pioneering software defined networking platform, enabling clients to rapidly create higher bandwidth virtual connections to Google Cloud and AWS Direct Connect, and between CoreSite data centers, including 50 gigabit services.
These upgrades reduce the time required for organizations to augment network capacity to support high bandwidth, low latency hybrid applications, like AI, machine learning, and digital media production. When it comes to current and future AI and machine learning applications, CoreSite's flexible, purpose built designed data centers, position us to host power intensive GPU services being used for AI and ML use cases.
For example, we're already providing GPU capacity for applications like 3D visualization and rendering and for software development within niche cloud environments. And for the dentist AI applications our purpose built facilities are designed to accommodate liquid cooling with modest development efforts when required. As we've stated previously, in the near-term, we continue to believe the majority of today's generative AI workloads will provide hyperscale opportunities that don't meet our investment criteria are fit within the CoreSite ecosystem.
However, as GenAI evolves, we would expect the balance of workloads to shift from large language model development, intensive training and public prompts to specialized inference based use cases as productivity gains from the deployment of custom models accelerates. At this age, when low latency interconnection, high power density and distributed high performance compute become the priorities, we believe CoreSite and ultimately, our distributed portfolio of franchise real estate assets across the US are going to be optimally positioned to benefit.
On that note, we've continued to see progress toward the realization of demand cases that support our initial edge thesis. And we believe we have an opportunity to enable a more efficient exchange of network traffic and support cloud services appearing in a more distributed manner. As a result, we've been working both internally and with external stakeholders to develop an edge model we can execute on is propelling opportunities present themselves and our initial assumption that through CoreSite our seat at the table and visibility into the customer demand environment would be materially enhanced is holding true.
We're increasingly seeing interest from potential customers looking to extend technologies, such as private cloud computing, AI, and 5G applications closer to the end device through more distributed architecture. This is resulting from several key demand cases including availability of future power requirements, business efficiency, revenue generation opportunities and customer experience.
When it comes to power, CoreSite has secured significant future power availability, and is insulated from expected shortages in markets like Northern Virginia. However, power constraints in general are increasingly in focus in legacy data center markets. In this case, we believe our distributed land footprint in tier two and three markets with significant power availability and capability to connect back to core side campuses can serve more distributed power capacity needs, while enabling customers to enjoy the interconnection benefits of the core site ecosystem.
And its potential customers increasingly focused on new revenue opportunities and customer experience, including through the proliferation of applications like next generation gaming, AR and devices and wearables that leverage interactive AI, we believe we have a compelling combination of distributed points of presence and interconnection capabilities that could be extended to a broader edge.
In addition, by prioritizing our existing owned real estate, which in many cases is already designated for uses digital infrastructure, we see an opportunity to drive a significant time to market advantage and reduce overall development costs, which could be compelling to customers and enhance returns on investment. As a result of these factors, we continue to work toward establishing a repeatable, rapidly deployable design with initial capacity and the one megawatt range, which could then be scalable to incremental megawatts with interconnection to multi site campuses is demand dictates.
As always, we'll assess potential growth at the edge through the prism of our discipline capital allocation framework, committing capital only if the opportunity meets our investment criteria and aligned with our long term strategic vision of growing our interconnection ecosystem in a way that maximizes shareholder value. In closing, the bottom line is that we remain at the relatively early stages of a 5G and network technology evolution that we believe will necessitate continuous incremental investment in existing infrastructure, like towers, data centers, and distributed edge infrastructure.
We also believe that ongoing technology developments will unlock new capabilities that will drive the next wave of innovative and data intensive consumer and enterprise devices and applications and American Tower with its leading tower portfolio and real estate footprint combined with a highly interconnected data center ecosystem is in a truly differentiated position to serve the network infrastructure needs of the future.
Before I hand the call over to Rod, I'd like to close my remarks by congratulating Steve Vondran who effective November one will hold the role of Global Chief Operating Officer until February one of next year, at which point he'll transition to the position of Chief Executive Officer and Bud Noel, who will become our new Executive Vice President and President of our US Tower Division.
The board and I have worked diligently on succession planning, weighing the merits of an external search against the talent we have within our organization. Steve joined American Tower in 2000 and currently serves as the Executive Vice President of our US and Canada business, including both towers and datacenters. For the past 23 years, Steve has been instrumental to the growth and sustainability of earnings for American Tower and his build tremendous credibility with our global organization, his peers, the Board of Directors, the American Tower investor base, and our customers.
All this to say, Steve is a clear candidate to lead American Tower in its next growth phase. And over the next several months, I'll work closely with Steve, the executive leadership team and the board to ensure a seamless transition. Lastly, I want to thank all of the incredible American Tower employees around the world, both past and present, our customers and investors for their support and confidence you've demonstrated since I joined in 2009. Although this is a difficult decision on my part, I look forward to the time ahead with family and friends and new challenges while watching the company under Steve's leadership, continue to succeed.
With that, I'll turn the call over to Rod to review our latest results and updated outlook. Rod?
