Listen to this story
After years of plummeting cloud computing prices, an upward trend in pricing was observed recently. The stark rise in cloud prices was such that Leon Kuperman, the chief technology officer at CAST AI, termed it as ‘cloudflation’.
Cloud prices have dropped consistently for quite some time now, driven by multiple factors such as price competition among the big three cloud service providers—Amazon, Google, Microsoft—rolling out of virtual machines equipped with new processors such as the Graviton series (AWS), EPYC (Advanced Micro Devices Inc.) and Altra (Ampere Computing Holding LLC) along with newer, more affordable chips such as Intel Corp x86 chips.
According to Synergy Research Group, enterprise spending on cloud infrastructure services continued to rise aggressively in 2020—growing by 35% to reach almost $130 billion.
(Image source: Synergy Research Group)
Infor—a multinational enterprise software company that makes business applications for aerospace, hospitality and other industries—observed that in 2018, its AWS bill grew 62% to about $130 million compared to the previous year.
In March 2022, Google announced pricing changes for Google Cloud Storage. The new pricing is scheduled to come into effect from October 2022 and is expected to entail rise in prices for several core storage features. For example, Nearline storage pricing in multi-region will increase by 50% from $0.010 per GB per month to $0.015 per GB per month. In addition, in regions, cold line storage Class A operations pricing will increase from $0.10 per 10,000 operations to $0.20 per 10,000 operations. Further, while default replication pricing in multi-region was previously free, it will now rise to USD 0.02 per GB for North America and Europe and USD 0.08 per GB for Asia.
AWS EC2 Spot Instance allows users to use spare Amazon EC2 computing capacity at a discount of almost 90 per cent on the on-demand price. However, Amazon offers this service with a caveat—AWS reserves the right to terminate and reclaim these spot instances with just two minutes’ notice. Given the uncertainty it entails, many cloud customers choose to stay away from spot instances.
Through this, Amazon came up with a creative customer strategy that keeps the prices floating at market rates—whenever there is an increase in demand for such cloud services, AWS increases the prices instead of terminating the service.
In 2022, these prices witnessed a notable rise.
What explains the increasing cloud costs?
The sudden rise in cloud prices after years of plummeting trend raised many eyebrows.
Disruption in semiconductor supply chain
The covid-19 pandemic, followed by the ongoing Russia-Ukraine war, significantly disrupted the supply chain of critical inputs essential to the cloud computing industry. For instance, ‘neon’ and ‘palladium’ used in semiconductor manufacturing are typically sourced from Russia and Ukraine. Such disruptions in the supply chain eventually led to scarce availability of servers in the cloud, resulting in hiked prices.
Uncertainty surrounding Taiwan
Taiwan is a major node for critical technological processes like chip manufacturing.
In 2021, Taiwan accounted for over 66 per cent share in the global chip-making market. Tech giants like Apple, Amazon, Google, Qualcomm and NVIDIA depend inordinately on Taiwanese contract manufacturers to produce up to 90% of their chips.
Considering the excessive dependence on Taiwan for semiconductors, it is considered an integral part of the manufacturing ecosystem by many tech-driven nations like the US and China. The control of the Taiwanese market translates to the control of a major source of innovation that drives computing power costs—a lucrative proposition universally.
The ongoing war between Russia and Ukraine has led to many resource-related issues. The response from other countries has further added to the problems—leading to an increase in energy prices across the globe but particularly in Europe.
Data centers are power intensive and rising energy prices are adding to their operating expenses. For instance, hyperscale cloud providers such as Google, Microsoft and Amazon have several data centers in Europe—‘Google Cloud’ in seven regions, ‘Microsoft Azure’ across 11 locations and ‘AWS’ in six regions.
Hyperscalers reach scale
Initially, as the user base expanded, the prices dropped due to optimisations, architectural improvements and economies of scale.
“As their customer base and usage grew from hundreds to millions of customers, cloud service providers drove optimisations from investments and shared resources (think mega datacenter operations). But these operational efficiency gains are now slowing down as hyperscalers actually reached that scale”, writes Laurent Gil, founder and Chief Product Officer at Cast AI.
Stalled expansion due to uncertain geopolitics
Big cloud providers have announced plans to expand their data centers in Europe—Microsoft announced plans to open data centers at six new locations in Europe; AWS is set to open a new data center in Spain and announced plans for a region in Switzerland; and Google announced plans to open four new centers in Europe.
New centers are deemed vital in order to accommodate the ever-increasing cloud consumption. However, the uncertain geopolitical landscape in Europe may stall these plans.
Tackling rising cloud cost
Most reasons cited above are beyond users’ control. Nevertheless, companies that are cloud service consumers can deal at their individual levels to tackle the ballooning cloud costs.
According to Laurent Gil, accurately forecasting cloud consumption costs is a daunting task. Even big companies with dedicated teams fail to forecast accurately.
The most reasonable way out in such scenarios is to reserve cloud resources while they are still cheap. Companies can work towards optimising their existing resources. Using specialised automation tools, they can select the right types and sizes of cloud resources; scale their resources up and down automatically to avoid overprovisioning and waste; and decommission resources that aren’t in use.
Cloud repatriation could also be a viable option.
Cloud repatriation is the shift away from the cloud to on-premise infrastructure. A study by venture capital firm ‘Andreessen Horowitz’ found that cloud repatriation results in one-third to one-half the cost of running equivalent workloads in the cloud. In the case of Dropbox, repatriation generated USD 75 million of savings over two years and gross margins went up from 33 % ( 2015 pre repatriation) to 67% in 2017.
In yet other instances, even simple steps like incentivising the right behaviour at the right place could help save substantial cloud costs. For example, providing spot bonuses to engineers who have saved a certain amount of cloud cost.
Other measures like diversifying the manufacturing base to avoid supply-side disruptions witnessed during the pandemic are also considered useful.
The cloud paradox
Cloud technology has done away with the need to spend heavily on infrastructural requirements such as servers and an expert team to handle such technology. Often, these are beyond one’s expertise and can detract them from their core area. In doing so, cloud technology has brought about a transformational shift in how businesses use technology. Software teams can now summon computing resources whenever needed and without hassle. Many companies found it easier and cost effective to pay for computing capacity to cloud service providers as per their needs. Certain cloud-specific features such as multitenancy (multiple customers simultaneously using shared servers and storage systems) increase the efficiency of applications, thereby keeping cloud costs down.
However, as companies scale and mature, they often lose track of their cloud consumption, and the cloud cost locks up billions of market cap—making it appear cheaper in the early stages but ultimately proving to be a costlier endeavour.