In the third quarter this year, the capital expenditure (capex) of hyperscale data center operators touched $26 billion. It’s the second time in 2018 that capex reached this level, according to the new data by Synergy Research.
The capex of hyperscale for the first three quarters in 2018 is up 53% as compared to the same period in 2017.
Hyperscale operators are the big companies which have a distributed infrastructure that can be scaled up to meet the increased demand for additional physical space, electrical power, cooling etc.
Top 5 hyperscale spenders
- Microsoft
- Apple
- Amazon
The research finds that Google, Microsoft, Facebook, Apple and Amazon are the top five companies to have spent the most on hyperscale infrastructure in Q3 2018.
These companies have been in the list of top five spenders for last ten quarters, accounting for 70% of hyperscale capex. The capex of Microsoft reached a record level in the third quarter, while the capex of other four leaders slightly decreased compared to the growth witnessed in last quarter.
Synergy mentioned that capex in third quarter would have been the highest ever, had it not been for Google’s purchase of Manhattan’s Chelsea Market building in March. This purchase had boosted the Q1 numbers by $2.4 billion.
Apart from these five, capex of Alibaba jumped up in Q3, putting the company way ahead of other hyperscale operators.
Following Alibaba, the other top hyperscale operators with highest capex included Baidu, IBM, JD.com, NTT, and Tencent.
“Business at the hyperscale operators is booming. Over the last four quarters their year-on-year revenue growth has averaged 24% and they are investing an ever-growing percentage of their revenues in capex,” said John Dinsdale, a Chief Analyst at Synergy Research Group.
“That is a real boon for data center technology vendors and for colocation/wholesale data center operators, but it has created a huge barrier for companies wishing to meaningfully compete with those hyperscale firms. This is a game of massive scale and only a few can play that game.”
Most of the spending on hyperscale goes towards building, expanding and equipping huge data centers. The number of these data centers have now reached 423 globally.
