For cloud providers, how IP addresses are used is an important indicator of their customer base and business growth. Although IPv6 has entered commercial use, the vast majority of users still rely on providers’ IPv4 addresses for global connectivity and cloud services. Yet many people have never deliberately paid attention to IP address usage.
1. Which cloud provider owns the largest pool of IPv4 addresses?
2. Which provider generates the highest revenue per active IPv4 address?
3. Are IP address utilization patterns really unrelated to a provider’s overall business performance and strategy?
With these questions in mind, this article takes an “IP address economics” perspective to examine the IPv4 holdings of the global cloud giants (Gartner’s IaaS Top 4: AWS, Azure, GCP, and Alibaba Cloud), and to observe their network infrastructure and market development from a different angle.
Before we dive into the analysis, it is useful to understand a few facts:
1. Global IPv4 address resources are essentially exhausted. In most cases, new requests are allocated IPv6 addresses only.
2. The current transaction price of a single IPv4 address is reported at around USD 20–30. Some organizations still hold sizeable IPv4 blocks that can be traded. For example, Japan’s WIDE project recently announced plans to release 14.68 million IPv4 addresses (87.5% of the 43/8 block), which is likely to drive prices down as telecom operators, internet companies, and cloud providers compete for these resources.
3. The distribution of IPv4 addresses is highly uneven across countries. The United States holds about 1.15 billion addresses (roughly a quarter of all allocatable IPv4s), while the three major US cloud providers together own about 120 million addresses—“richer than some nations”, given that Germany has about 110 million.
The analysis that follows reflects personal technical views and is intended only as a discussion of industry phenomena. Any inaccuracies are welcome to be corrected. IP‑related data in this article is based on the “Global Cloud Service Providers 2019 1H IP Address Analysis Report” published by IPIP (Tiantexin), to whom we give special thanks.
All five Regional Internet Registries (RIRs) under IANA have stated that their IPv4 pools are effectively depleted. Large new allocations of IPv4 addresses have stopped; only small blocks such as /23 are occasionally issued based on each RIR’s remaining reserves and reclaimed space, typically prioritizing new members.
Charts of IPv4 and IPv6 holdings by country (not reproduced here) show that the US has about 1.15 billion IPv4 addresses and China about 346 million.
Because IPv4 allocation over the decades has been extremely uneven, once the RIRs stopped issuing large new blocks, market demand and profit motives quickly drove many organizations—especially those that had obtained large legacy allocations—to sell unused address space. Addresses that closely match a buyer’s region and routing needs tend to command higher prices, while large cross‑region blocks see more flexible pricing.
From a basic economics perspective, in a market for a scarce resource, “first movers” and “highest bidders” ultimately win. IPv4 prices have accordingly risen from free, to cents, to dollars, and now to tens of dollars per address. Until IPv6 becomes the dominant protocol for global networks and applications (optimistically in 5–10 years), IPv4 will remain a strategic resource. As more large holders release significant blocks into the market, a new wave of IPv4 trading is likely, with prices following market forces.
For cloud providers, IP addresses are strategic assets. Although RIRs largely stopped issuing new IPv4 blocks around 2012, providers have continued to meet growing customer demand through a combination of local and cross‑region IP trading and optimization of address usage—effectively “opening new sources and reducing waste” in their IPv4 pools.
Because their consumption is so large, cloud providers typically obtain addresses directly from RIRs such as ARIN and APNIC, or through third‑party transactions later registered with the relevant RIRs.
As of the end of June 2019, AWS leads the world with 75.24 million registered IPv4 addresses. Azure, Alibaba Cloud, and GCP have 38.14 million, 15.52 million, and 8.87 million respectively.
Notably, AWS has aggressively stockpiled IPv4 addresses since 2017. From 2017 H1 to 2019 H1, its registered pool increased by 44.06 million addresses. This reflects a combination of strategic foresight and large‑scale address acquisitions. Over the same period, Azure, Alibaba Cloud, and GCP increased their pools by 14.89 million, 5.51 million, and 7.85 million respectively.
There are also significant differences in the ratio of active to registered IPv4 addresses. Thanks to its stockpiling strategy, AWS’s active‑address ratio has fallen to 14.93%, leaving ample room for future growth. GCP’s active‑address ratio, however, remains as high as 41% (down 21.2 percentage points from 2017 H1), suggesting very rapid business expansion (active addresses jumped from 630,000 to 3.67 million) and possibly a greater focus on IPv6, with less emphasis on building a large IPv4 buffer.
If we could further break down where each provider’s registered and active IPv4 addresses are located, we would gain an even clearer understanding of their infrastructure priorities and regional strengths.
Active IPv4 Usage by Region
At a high level, active IPv4 usage by region can reveal a provider’s customer growth trajectory, strategic focus, and market position:
–AWS: Most active IPv4s are in the US and Europe. Active users in the US account for 54.06%; Ireland 8.44%; Japan 7.69%. Active users in mainland China and Southeast Asia together are less than 1%.
–Azure: Active IPv4s are also concentrated in the US and Europe. In the US, active users account for 48.15%; the Netherlands 18.79%; Ireland 7.81%. Hong Kong accounts for 2.59%; mainland China effectively has no active IPv4s; Singapore 3.91%; India 1.83%.
–GCP: Active IPv4s are mainly in North America and Europe. In the US, active users account for 58.83%; Belgium 10.44%; Taiwan 6.18%; Hong Kong 1.79%; Singapore 1.91%.
–Alibaba Cloud: Active IPv4s are concentrated in Greater China (Beijing, Zhejiang, Guangdong, Shanghai, Hong Kong, Shandong, etc.), with over 80% in the region. The US and Singapore each account for about 2%; there is essentially no footprint in Europe.
The regional active‑address ratio is calculated as: active IPv4s in the region / total active IPv4s for that provider.
The regional breakdown of active IPv4s clearly shows that AWS, Azure, and GCP have most of their customers and workloads in Europe and North America, while Alibaba Cloud is concentrated in China.
We also see an interesting pattern: the distribution of active IPv4s broadly correlates (though not linearly) with each cloud provider’s global network infrastructure deployment (regions and AZs). AWS has built deep infrastructure in key regions in the Americas and Europe (the US and Ireland), gradually expanding to more European and Asia‑Pacific regions (such as Japan). Azure focuses on the US and the Netherlands, with the broadest geographic footprint. GCP leverages its global backbone and 100,000 km of submarine cables to prioritize the US, Belgium, Taiwan, and other regions. Alibaba Cloud, while globally deployed, still centers its core infrastructure around China and the surrounding Asia‑Pacific markets.
These patterns match what we see in each provider’s real‑world business footprint and infrastructure maps.
Looking at active IPv4 market share in key regions (such as the US, China, and Southeast Asia), we find that these markets are highly concentrated: the top three cloud providers account for more than 90% of active IPv4s. AWS clearly leads in the US, Europe, and Southeast Asia, while Alibaba Cloud leads only in China.
Here, the regional active‑address market share is calculated as: a provider’s active IPv4s in that region / total active IPv4s in that region.
Putting active IPv4 share side by side with revenue share, we observe a general pattern: providers with higher active IPv4 share tend to also have higher revenue share. However, Azure stands out as an exception: its active IPv4 share is only 8.5%, but its revenue share reaches 16.9%.
If we normalize revenue by active IPv4 address, Azure has ranked first in the past two years (even though its 2019 figure is slightly lower than 2018). Its revenue per active IPv4 is more than twice that of AWS, indicating that Azure achieves the best “revenue per address” among the major players. GCP is again a counterexample: its revenue per active IPv4 has dropped significantly (from 0.52 to 0.17). One possible explanation is that over the past two years GCP has aggressively attracted developers and startups, causing active IPv4 usage to skyrocket (from 630,000 to 3.67 million), while revenue has not grown at the same pace. Alibaba Cloud’s metric is relatively stable, suggesting its address consumption and revenue growth have remained in sync.
We can tentatively draw two conclusions:
1) Cloud providers with higher revenue per active IPv4 generally perform better in revenue and market share (AWS and Azure rank first and second globally).
2) Providers with higher revenue per active IPv4 can generate more cloud revenue with fewer IP address resources, giving them stronger long‑term sustainability.
Cloud providers should closely monitor changes in revenue per active IPv4 and analyze them together with active‑address growth and revenue growth. By understanding the underlying drivers, they can aim for best‑in‑class efficiency—achieving strong growth while consuming as few IPv4 addresses and related resources as possible.
In this article we used a simplified “IP address economics” framework to observe and analyze how IP addresses reflect the development trajectories and market performance of cloud providers. In practice, deeper IP address analytics can help all parties in the ecosystem—especially cloud providers—optimize cloud network infrastructure, internet routing, and user experience.
With more granular IP address data combined with solid statistical methods, this kind of analysis could provide consulting firms with a fresh perspective for evaluating cloud providers, helping reports such as Gartner’s Magic Quadrant better reflect the realities of the cloud market.
As IPv6‑only customers and workloads become more widespread, the methods described here will need to be extended to include IPv6, and current conclusions may need to be revisited in light of new data.
Technical support
Scan the code to join the technical support WeChat group
Official account
Scan the code to follow and get the latest updates