Where to begin with this.

First off, unless you have independent channels into either the CPU or directly to memory, this won't be fast(er). Second, in storage, latency is the dominant factor, not capacity or transfer speed for most applications ( except for archival and similar write once applications ).

RAID ( correctly identified here is Redundant Array of Inexpensive Disks - later named by the drive makers as Redundant Array of Independent Drives for obvious reasons ) was actually invented in the late 1960s... Drum storage was, in fact, a type of Raid 0 or RAID 2 depending. What Garth Gibson ( author of the 1989 paper on RAID ) did at UC Berkeley was to come up with a classification of existing RAID systems and, most importantly, develop a failure analysis mathematics that described the application of redundancy. Finally his group at UC Berkeley developed RAID 5, a variation of RAID 3 ( or block parallel RAID with single XOR parity drive ).

But it was important that each drive have independent path to the CPU ( or DMA access to memory ).

I sort of know about all this because I caused NASA to fund Dave Patterson, Randy Katz, and Garth Gibson ( two sponsoring professors and Garth, the grad student ). Not only that, but I got early access to the seminal RAID paper before it was published in JACM ( Named a "Taxonomy of Redundant Arrays of Inexpensive Disks" ). I used that access to write ( for NASA ) the very first RAID 5 implementation anywhere in the world for Amdahl UTS... using channel programming on our Amdahl to do some of the RAID operations thus offloading the CPU. I wrote a paper on this and presented it at another ACM conference for Unix... and to my horror I had Dennis Ritchie sitting in the first row... and I about shit my pants when I gave my talk. Look up Dennis Ritchie on google to see his contributions to the world of computing.

Anyway, Garth's paper spawned an industry... I remember all sorts of start up NAS companies visiting NASA and getting me to "teach them about this RAID concept", including people that would go on to found companies like DDN, NetApp, etc.

The reason for RAID was two fold, redundancy so that a single failure would not lose your data ( or even lose access to your data ) and because while processor and memory were rapidly increasing in speed, mechanical disk drives were not keeping pace. The only way to move data more rapidly from non-volatile storage to computer memory was by going parallel. Later, processors gave up the ghost on executing with ever faster clock speeds and also went parallel ( more cores ) resulting in multicore CPUs... and limited instruction set massively parallel GPUs ( first for video games but now for AI ). Storage speeds are still an issue. When I was at Seagate later in my career, I led a team that developed a parallel file system on top of RAID storage nodes that moved data at 1 TB ( TeraByte ) per second... that was in 2012.

Anyway, this was rather a silly idea... sort of like building a motorcycle out of 8 V-8 car engines. Yes you can do it, maybe even get people to look at what you did, but it isn't useful and everyone will think it looks ridiculous. Which it does.