WiFi 7's MLO Feature Is About Reliability, Not

The TP-Link EAP787 promises 12,000 Mbps of aggregate throughput across three wireless bands, with the kind of specs that make networking enthusiasts salivate: 320MHz channels in the 6GHz band, 10 gigabit Ethernet uplink, and Multi-Link Operation for WiFi 7 clients. YouTuber apalrd spent his own money upgrading test equipment to properly evaluate these features, and what he found illuminates a fundamental gap between how WiFi 7 is marketed and how it actually works.

The headline feature—Multi-Link Operation—turns out to be substantially different from what most people imagine.

The MLO Misconception

When WiFi 7 vendors tout Multi-Link Operation, the natural assumption is that your device will simultaneously transmit on both 5GHz and 6GHz bands, essentially doubling your bandwidth. That's technically possible under the standard. It's just not what happens in practice.

"When most people think of MLO, they're like, 'Wow, my device can use 5 GHz and 6 GHz at the same time to send like what that be 240 MHz in the 5 GHz band plus 320 MHz and 6 GHz band like 10 Gbits per second or something,'" apalrd explains. "And while the standard allows that to be implemented, no actual radios that I can find implement it."

The reality is more modest: most WiFi 7 client devices—including the ubiquitous Intel BE200 chipset that's showing up in new laptops—support only Enhanced Multi-Link Single Radio (EMSLR). This splits the device's two antennas so one listens on 6GHz and one on 5GHz, but transmission still happens on just one band at a time. The access point or client can dynamically switch between frequencies on a frame-by-frame basis, but not transmit simultaneously on both.

That's not bandwidth aggregation. That's interference avoidance.

"So that's not going to get you 10 Gbits per second," apalrd notes. "It's going to get you a more reliable gigabit, which for most people is a fantastic thing, but it's very misunderstood, I think."

What the Hardware Actually Delivers

In real-world testing with multiple clients across different bands, apalrd achieved roughly 3 Gbps aggregate throughput—impressive, but nowhere near the theoretical 12,000 Mbps ceiling. A single WiFi 7 client on the 6GHz band peaked at about 2 Gbps, which drops to around 1.3 Gbps as a sustained average. WiFi 6 clients on the 5GHz band managed around 1 Gbps.

These numbers reveal why so many consumer access points ship with 2.5 gigabit uplinks rather than 10 gigabit. For typical home deployments, even with multiple simultaneous clients, you're unlikely to saturate a 2.5 Gbps connection. The EAP787's 10 gigabit uplink positions it squarely in enterprise territory—which makes sense given its price point and feature set.

The testing also exposed practical quirks. The access point runs hot enough to be noticeably warm to the touch during sustained load. Band steering—getting clients to connect to 6GHz instead of 5GHz—proved finicky, with devices sometimes requiring multiple connection attempts to land on the optimal band. The 6GHz scanning process involves a two-step dance where clients first check 2.4GHz and 5GHz to determine their regulatory region before probing 6GHz channels, which can slow initial connections.

The Standards Timing Problem

One detail that often gets lost in WiFi 7 hype cycles: the 802.11be standard only officially released in September 2024. Yet consumer manufacturers started shipping "WiFi 7" products based on draft specifications more than a year earlier.

"In reality, consumer manufacturers started selling like cheap consumer access points that claimed Wi-Fi 7 based on the draft spec like a year and a half before the spec even came out," apalrd observes. "It's like, how can you make something that's not even standardized yet?"

Both TP-Link and Ubiquiti waited for the finalized standard before shipping WiFi 7 products—a conservative approach that earned them criticism for being "late to the game" while arguably being more responsible. The WiFi Alliance's willingness to let manufacturers ship pre-standard hardware created a market dynamic where patience looked like slowness.

The Controller Flexibility

One genuinely differentiating aspect of TP-Link's Omada ecosystem: you're not forced into a specific management model. The EAP787 can run standalone with its built-in web interface, connect to TP-Link's free cloud controller for small deployments, use a hardware controller, or run with a self-hosted controller instance.

That last option is what apalrd tested, running the controller software in a Debian 12 LXC container. The controller is Java-based and memory-hungry—consuming 2-3GB and prone to running out of memory if not given enough headroom—but it provides legitimate enterprise features including IPv6 support, VLAN configuration, and centralized management without forcing you into vendor lock-in.

"This is different from some other vendors where you must have a controller, like say Ubiquiti," apalrd notes. The comparison is pointed: Ubiquiti's ecosystem generally requires their controller software even for basic deployment, while TP-Link offers genuine flexibility.

Should You Wait?

The honest answer for most people: yes. Not because WiFi 7 is bad, but because the client device ecosystem isn't ready. Apalrd had to specifically purchase new hardware to test WiFi 7 features—none of his existing devices supported it, despite some being only a couple years old.

WiFi 6 devices will continue working fine for years. The reliability benefits of MLO are real but incremental. The bandwidth improvements only matter if you're regularly moving multi-gigabit files over wireless, which remains a niche use case. And the premium pricing on WiFi 7 hardware hasn't yet been justified by mass-market client adoption.

For enterprise deployments with specific high-density or high-bandwidth requirements, the EAP787 makes sense now. For home users eyeing those BE12000 numbers, the smarter play is waiting until your actual devices support WiFi 7—not just the spec sheet, but the reality of what that support delivers.

Because bigger numbers aren't always better. Sometimes they're just bigger.

—Marcus Chen-Ramirez