Twitter's new owner and CEO Elon Musk is debating whether to shut down one of Twitter's three main US data centers.
The New York Times reports that the Sacramento 'SMF1' data center could be on the chopping block, according to four people with knowledge of the idea.
Should it be shut down, Twitter would be left with data centers in Atlanta and Portland, Oregon. It also has cloud contracts with Amazon Web Services and Google Cloud.
But Musk is believed to also be looking to trim the company's cloud costs, aiming for overall infrastructure savings of $1 billion a year. His $44bn acquisition added new debt to the company, which coincidentally increased Twitter's interest repayments to $1bn a year.
After taking over the business, Musk fired roughly half of the company's 7,500 staff. Soon after, a server room at its headquarters overheated as no one was left to maintain it.
He then fired people that criticized him publicly, as well as searched through private company messages for critics to lay off.
Then he gave remaining workers a deadline to decide whether to stay, offering severance packages to those who left and promising "long hours at high intensity" and an "extremely hardcore" environment for those that remained.
It appears more than a thousand Twitter employees quit - but no one is exactly sure how many left, as communications and human resources departments have also been decimated.
The Twitter Command Center, a 20-person team tasked with preventing outages and tech failures during periods of high demand, lost numerous people, the NYT reports. The 'core services' computing architecture team went from more than 100 people to four.
Other teams fell by more than 80 percent, while some are believed to be completely empty.
The sudden cuts have led to concerns about glitches and outages in the days and weeks ahead - compounded by the start of the World Cup, which is expected to lead to high traffic periods.
Twitter has experienced numerous outages before - including in September, when the Sacramento data center collapsed amid an extreme heat wave - but historically had staff and data center capacity on hand to restore services.