Fundamentals
A brief history of the MAC address
The 48-bit format on every network interface today traces back to a lab at Xerox PARC in 1973, and it's been extended, not replaced, ever since.
1973 to 1980: Ethernet, before it had a number
Ethernet started at Xerox PARC, designed by Robert Metcalfe and David Boggs as a way to link the lab's early workstations and printers over a shared cable. An addressing scheme was part of the design from early on: something to tell devices on the shared wire apart without a central coordinator assigning numbers by hand. That scheme was refined through the DEC-Intel-Xerox partnership and published as "DIX Ethernet" (the Ethernet Blue Book) in 1980, already using a 48-bit address.
1980 onward: IEEE standardization, and the OUI registry
1980 is also when IEEE formed Project 802 to standardize local area networking more broadly, not just Ethernet. IEEE 802.3 (Ethernet) was published in 1983 based on the DIX work, and other 802 standards that followed, Token Ring, and later Wi-Fi (802.11), adopted the same 48-bit address format for consistency, which is why a Wi-Fi card and an Ethernet port use addresses that look identical and can even collide in the same address space if you're not careful.
Making that shared 48-bit space collision-free without a central authority handing out every individual address was the actual hard problem, and the answer was the OUI: IEEE keeps a registry (still active today) assigning each manufacturer a block of addresses, identified by the first three bytes, that only they may use. A vendor picks addresses within their own block however they like; two different vendors' blocks never overlap. How MAC addresses are structured covers exactly how that split works today, OUI blocks and all.
Extending the format: EUI-64 and IPv6
IEEE later defined EUI-64, a 64-bit sibling format for identifier spaces that needed more room (FireWire, ZigBee, and a few other technologies), and specified a standard way to expand a 48-bit MAC address into a "modified EUI-64" by splitting it and inserting a fixed FFFE in the middle. IPv6 adopted that exact conversion for stateless address autoconfiguration: an interface's IPv6 link-local address can be derived directly from its MAC address, no DHCP involved. Our lookup tool performs this derivation live for any address you enter.
2017: IEEE 802c and the locally administered space
For decades, "locally administered" (the U/L bit set to 1) was a loosely defined catch-all: anything not drawn from an IEEE-assigned OUI. IEEE 802c, published in 2017, formalized that space into a structured local address plan (SLAP), carving it into defined quadrants for different assignment purposes rather than leaving it a free-for-all. It's the standard that gives today's locally administered addresses, including randomized ones, an actual defined structure instead of just meaning "not a real vendor address."
2014 to today: randomization
The most recent chapter isn't a new standard so much as a new use for the old locally-administered bit: starting with iOS 8 in 2014, mobile operating systems began generating a fresh, randomized, locally administered MAC address for Wi-Fi scanning instead of broadcasting the real hardware one, specifically to stop passive tracking by anyone with a Wi-Fi scanner. How MAC randomization works covers the mechanics; from a history standpoint, it's the moment the "locally administered" bit went from a niche flag mostly seen on virtual machines to something a majority of phones set by default, every single day.
Fifty years on from Xerox PARC, the address format itself hasn't changed. What's changed is who's assigning it, and why.
More guides