Apple confirms $1.3B Iowa data center, says operations will start in 2020
As anticipated, Apple on Thursday confirmed plans to build a new data center in Waukee, Iowa, which it says will cost $1.3 billion, plus up to $100 million more for a "Public Improvement Fund."
The complex, near Des Moines, will span 400,000 square feet, and handle services like Siri, iMessage, and the App Store, Apple said. Power will stem entirely from renewable energy, including "wind and other sources."
The Public Improvement Fund will be managed by the City of Waukee and develop "community projects like parks, libraries and recreational spaces, as well as infrastructure needs." The first planned effort is construction of the Waukee Youth Sports Campus, which will include fields, a greenhouse, a playground, and even a fishing pier.
It should take some time for the Fund to hit the $100 million mark, as the company will only be paying in about $1 million per year.
Similarly, though Apple indicated the data center should create "over 550 construction and operations jobs," it's likely that just 50 of those will be permanent. Apple is receiving some $213 million in incentives from state and local government.
Construction should start in early 2018, and bring the center online by 2020.
Apple CEO Tim Cook is slated to appear in Des Moines for the announcement, and will visit students at the Waukee Innovation and Learning Center alongside Governor Kim Reynolds.
The complex, near Des Moines, will span 400,000 square feet, and handle services like Siri, iMessage, and the App Store, Apple said. Power will stem entirely from renewable energy, including "wind and other sources."
The Public Improvement Fund will be managed by the City of Waukee and develop "community projects like parks, libraries and recreational spaces, as well as infrastructure needs." The first planned effort is construction of the Waukee Youth Sports Campus, which will include fields, a greenhouse, a playground, and even a fishing pier.
It should take some time for the Fund to hit the $100 million mark, as the company will only be paying in about $1 million per year.
Similarly, though Apple indicated the data center should create "over 550 construction and operations jobs," it's likely that just 50 of those will be permanent. Apple is receiving some $213 million in incentives from state and local government.
Construction should start in early 2018, and bring the center online by 2020.
Apple CEO Tim Cook is slated to appear in Des Moines for the announcement, and will visit students at the Waukee Innovation and Learning Center alongside Governor Kim Reynolds.
Comments
A $100 million fund, but Apple pays only $1 million per year? Is that a joke? So Iowa can pay off the public facilities they're going to build in 2117?
And it's costing Apple $3250 per square foot to build the place (including land costs I would assume). That sounds kind of costly to me considering it's not much more than a warehouse with a lot of air conditioning.
Just in case it's the latter:
On it's own, Apple's cloud services division is a Fortune 100 company.
On its own, Apple's services division makes more dosh than the whole of Facebook.
In 2014, the service was peaking at 200,000 messages. per second. That was in 2014. The service sends 40 billion notifications a day.
Videos, photos, FaceTime, music, app delivery, storage…
Next time, remember the smiley, or people might think you're clueless.
All of the data centers will also help Apple as its services continue to expand for AR/VR, AI, Music, video, Siri, Maps, autonomous driving, etc.
Great to finally meet an expert here!
So that last energy-efficient, high-reliability, ultra-scale data centre you built… How much did it cost?
iMessage. FaceTime. Siri. App Store. iTunes Store... I could go on, but you get the idea.
Also, this is Apple planning for the future. Think about that for a few hours...
You must be new to the Apple scene? Apple has been offering cloud services for 16+ years, and their global computing needs have only increased 100-fold in the past 10 years. For many years, Apple _did_ use other providers, such as Amazon and Google, but they are a company that wants control over everything, so they are building their own data centres now.
Geographic proximity to users is a very important factor. The closer the data center, the better the performance of all services offered. Geographic diversity also lowers the impact during unexpected outages.
40 billion per DAY? That's 463,000 per SECOND!
I remember when Apple was taking their time building their notification platform, and they did not want to release it until it could truly scale. Guess they really did achieve that. Notifications work instantly and perfectly for me.
I wonder if that 40 billion a day number includes all notifications including those that are entirely local (a notification generated by an app on your phone) that never hit Apple's infrastructure. But even if half of those are local, it's an enormous number.
https://www.wired.com/2017/06/pied-pipers-new-internet-isnt-just-possible-almost/
Bittorrent is a widely used example of this kind of system but that still relies on HTTP and search engines to find the nodes. The DNS system can be decentralized with a blockchain. The main ledger doesn't need to point directly to files but to nodes with those references, which keeps the sizes of the ledgers manageable and scalable, which Bitcoin should really do (e.g maximum 1m transactions per ledger and hash the ledgers by timestamp or id range).
The transfers can be encrypted end-to-end. This is how IPFS does it. Every client has a key pair, they can broadcast the public key so when they retrieve a file, the block is encrypted in a way only they can decode it so during transmission, there's just a part of a file that is encrypted to every user.
The battery and network capacity on mobile phones are limitations just now but not PCs. Netflix has ~5000 servers and tens of petabytes of storage (10,000 shows x 2GB = 20PB). To store this with redundancy, say 5 copies per file is 100PB of capacity. 5GB per device = 40m users. Eventually consumer devices will have multi-TB SSDs so it could reserve as much as 1TB per user. It would be best to ensure no host ever has a full copy of a file, the files should be split in uneven blocks to avoid legal issues and to allow scaling. There needs to be caching nodes too for high demand content so that 20 million people aren't trying to get a block from 1 device at the same time, when the capacity exceeds a demand (e.g 20 requests for 1MB each), the blocks would be distributed exponentially through the cache nodes and served that way.
This kind of network can replace user-driven services like twitter, Facebook and Youtube. To some extent Google as the content can be indexed by the lower nodes. It couldn't replace authoritative services because people need to trust the source.
It could replace the ad system for some uses too. If people pay for backup storage, there can be news feeds where the agreement is that on viewing the content, the reader's device is used as a cache/storage node and the news provider gets paid based on how many nodes they get.
It's a vulnerable system because it needs devices to be left on. At night, people shut devices off. Servers run 24/7. The demand in certain regions goes down too but that's no good for a global system. The file blocks can be split geographically to ensure more redundancy. I think for user-generated content, especially low storage things like chats (Skype, Twitter, WhatsApp, SnapChat, websites), it can be a good alternative to mainstream internet services.
I don't see it ever completely replacing central server operations because of the need for companies to control their content and be able to take it down or change it when they want to. For example if Apple no longer has a license to distribute music or movies on their service, the network can't keep serving it outside of their control.
The costs to run a cloud service are going to fall. The population of the world probably isn't going to double in 100 years and SSDs can scale to hundreds of TB per drive (100TB this year):
https://www.geek.com/chips/toshiba-hard-drives-will-be-40tb-by-2020-ssds-will-be-128tb-by-2018-1632425/
Server chips will be able to scale another few multiples from now and the rate of improvement in servers should outpace the demand. This is a good reason for Apple not to get back into the server business. Apple's not going to keep needing new data centers forever, once they can manage the load, they will just maintain the ones they have. I doubt they'll need more than 10-15 data centers. This would give them ~1m servers to handle 1.5b users.