WavKong V2700: $520, all network ports are Gigabit.
On the morning of November 1, while browsing WavKong's official website, I saw a router listed at USD 520, with a crossed-out original price of 703 next to it. In the Chinese context, this number is a bit jarring—the Wi-Fi 6 routers you could buy on e-commerce the same day cost only a fraction of it. So the first reaction can only be: what justifies it.
Let's clarify the timeline first. WavKong is a consumer brand under Digital Enhancement; its product launched on Kickstarter on April 22, 2026, drew 1,074 backers and raised HKD 3,272,087, shipped the final batch on September 29, and on October 30 announced commercial availability via a PR Newswire release in New York. In other words, this is not a new product from a major vendor, but a small company that used crowdfunding to build its own chip. When evaluating it, this has to be taken into account.
Official specs, reproduced as-is
The platform is listed as MTK7981 AX3000, with 128MB flash, 256MB memory, and 2×2 antennas each for 2.4GHz and 5GHz. The Ethernet ports are 1 Gigabit WAN plus 3 Gigabit LAN, all Gigabit, with no 2.5G. Wireless is Wi-Fi 6, dual-band, supporting 160MHz bandwidth and 1024-QAM, encryption up to WPA3, with no 6GHz band. Power is 12V/4A; the body is 281×166×79 mm and 1.32 kg. Operating modes provided are router, bridge, and AP+client; cloud infrastructure is in the AWS North America region.
The official pitch makes three claims: a single unit covers up to 15,000 sq ft (indoors and outdoors combined), is 3 to 10 times faster at medium and long range, and can connect more than 100 devices. We'll break these down separately later.
This chip is used in hundred-yuan devices in China
MT7981 is not new. It is MediaTek Filogic 820, officially positioned as the entry-level Wi-Fi 6 SoC in the Filogic series, with a dual-core Cortex-A53 running at 1.3GHz and a 12nm process. In public documentation, OpenWrt's official reference board OpenWrt One uses it, and models available in China such as Xiaomi AX3000T and GL.iNet MT3000 also use it.
The memory configuration is even more telling. Xiaomi's official website lists the AX3000T as: dual-core 1.3GHz, 12nm, 256MB memory, and officially claims this 256MB memory can stably connect up to 128 devices (lab conditions, 2.4G/5G/wired combined). The V2700 has 256MB memory plus 128MB flash, and officially says it can connect more than 100 devices.
In other words, the "256MB memory + over a hundred devices" claim is an official statement that, in China, a machine commonly sold on JD.com for 129 yuan can already provide (page on November 1, subject to official current price). So the USD 520 is not buying this SoC, nor this memory configuration.
Also worth mentioning: MT7981 natively supports 2.5G Ethernet (HSGMII interface), and implementing a 2.5G port does not require an additional conversion chip. The V2700 only provides Gigabit, which is a product-definition choice, not a chip limitation.
So where is the USD 520 spent: on a chip that corrects the signal
Officially it is called RPU, Radio Processing Unit, an in-house ASIC developed over six years. What it does is called digital pre-distortion, DPD.
This has been used in base stations and satellite communications for decades. When a power amplifier operates near full power, it enters the nonlinear region; the output waveform is no longer proportional to the input, and points on the constellation are pushed off and squashed. Once the distortion reaches a certain level, the receiver can no longer distinguish adjacent points and can only fall back to lower-order modulation—the rate drops by a step, and it drops off a cliff, not just slows down linearly.
DPD works by "pre-distorting": before the signal enters the PA, it applies inverse compensation based on the PA's distortion characteristics, so that after the PA distorts it, the output comes out clean. The principle is not complicated; the hard part is computing fast, computing power-efficiently, and fitting it into the cost of a consumer-grade chip.
The official lab data is: under 160MHz bandwidth and 1024-QAM, EVM improves from about −29.3 dB to −41.2 dB, an improvement of more than 12 dB.
Why these 12 dB are worth calling out separately
EVM is error vector magnitude, measuring how far actual signal points deviate from ideal positions; the more negative the value, the better. It is tightly tied to modulation order: the higher the order, the closer the constellation points are, and the stricter the EVM requirement.
IEEE 802.11ax requires EVM no higher than −35 dB for 1024-QAM (measured with amplitude drift compensation disabled). This can be found in LitePoint technical documents and several RF test materials; it is not something I derived.
Plug the thresholds in: −29.3 dB does not clear the bar, while −41.2 dB clears it with margin. So what this set of official tests really shows is—Without DPD, this machine cannot sustain 1024-QAM at high power; only with DPD does it dare to turn it on.
This understanding is far more useful than "12 dB improvement": it is not a nice-to-have optimization, but a switch for "whether the highest modulation tier can be used." Of course, the lab is not the living room—what power amplifier, how much power, what temperature, none are disclosed by the official source; defer to the vendor's official information, and only third-party retesting counts.
But DPD fixes only one direction
This is the one sentence I most want to say after reading the material.
Wi-Fi is bidirectional. The router cleans up its transmit side, but the transmit side of your phone and laptop is unchanged. Downlink (router to you) benefits; uplink (you to router) does not.
In reality, what often breaks first is exactly the uplink. Video calls stutter, often not because downlink is insufficient, but because your voice and video cannot get out; home cameras keep dropping offline because the path they upload on cannot hold up; backing up photos to NAS and uploading large files to online storage are all uplink. This kind of traffic also has a hidden amplifier: TCP relies on uplink ACKs for confirmation, and once uplink is congested, the downlink window shrinks with it—speed-test numbers may still look fine, but actual use stutters.
That official "123 Mbps at the garden test point versus 4.2 Mbps for Mesh" measures downlink. I believe the gap is that large because that is an extreme location—but interpreting it as "all devices in the home are 30 times faster" goes too far.
The house used for testing is the kind most favorable to it
The official press release states it clearly: a single-story, wood-frame residence in California, and tests were also done in three-story wood-frame houses in other US states. The figure of 15,000 平方英尺 was obtained under such wall conditions.
Interior walls in North American homes are drywall plus wood studs, with cavities in between, so 5GHz attenuation through one or two walls is limited. Most homes in China are brick-concrete or reinforced-concrete load-bearing walls, and the attenuation of the steel mesh to 5GHz is on another order of magnitude. The same machine that can cover an entire floor in a single-story wood house in California may need repositioning in an ordinary three-bedroom apartment in Fuzhou.
So when you see "covers 15,000 平方英尺", first think about what walls that number was measured through. The official source itself also notes that "actual results depend on walls, building materials, interference, placement, client capabilities, and network configuration."
Still selling Wi-Fi 6 in November 2026—the official source explains it itself
The third question in the official FAQ is "Why not use Wi-Fi 7?" The gist of the answer is: this machine is centered on coverage and usable speed, not peak values measured right next to the router; Wi-Fi 6 compatibility is more mature.
I think this explanation holds up halfway.
In terms of whether it is enough, Wi-Fi 6's 160MHz plus 1024-QAM is indeed sufficient below gigabit broadband. Wi-Fi 7 in China has consistently had low cost-effectiveness, largely because the 6GHz band has not been opened, and China-market models lack the most valuable band (that article on whether you should buy Wi-Fi 7has broken it down). So the judgment that "there is no need to chase Wi-Fi 7" actually holds even more for users in China.
But 520 美元 is another matter. This price point is not a "good enough is fine" price point; it is a "I want to be better than others at the same price" price point. Fighting at this price point with Wi-Fi 6 and relying only on one RPU is too thin.
Four Gigabit ports means NAS users are shut out.
This is the most practical point, and the main reason I think this machine has little relevance to most readers of this site.
In China, many people willing to spend more than 500 yuan on a router are aiming for a 2.5G LAN: 2.5G output from the ONT/optical modem, 2.5G input to the router, distribution via a switch, and NAS and desktop machines fully saturating it. The barrier to this setup is not the router itself, but that every link cannot be all gigabit (I, inthat 2.5G LAN articlebroke it down by the three-piece set of switch, Ethernet cable, and NIC).
V2700 has 1 gigabit WAN plus 3 gigabit LAN ports. If you have a 2.5G NAS and a 2000M broadband line, connect them and they directly downgrade to gigabit, and SMB file copies drop from over 200 MB/s back to just over 100 MB/s.
The people it wants to serve are clearly written: large houses in North America, wood-frame construction, and not wanting to put three Mesh nodes in the living room. These are two different groups from those in China tinkering with 2.5G LAN. Like last month's 588-yuanHuawei router X2 Proalike, the money is all spent outside the spec sheet—with that one, you're paying for placement and styling; with this one, you're paying for RF.
As for the claim of "installing a few fewer nodes," I agree with half of it. The biggest problem with Mesh has never been the technology; it is placement: nodes placed at the end of a hallway and chained across three floors via wireless backhaul cannot be saved by two more units (Where Mesh Should Be Placedthat article did the math). A router with a clean transmitter may indeed let you buy one fewer node. But if the house has concrete load-bearing walls and the floor plan is chopped into fragments, link improvements cannot solve blockage, and you still have to add nodes when needed.
Three types of people, three answers
People with a NAS at home and building a 2.5G LAN: skip it. Four gigabit ports alone are enough to rule it out, and adding the $520 price and overseas after-sales support, it fails all three.
Large homes, people tired of being troubled by Mesh, and those who don't want to tinker anymore: worth watching, but don't rush to order. Wait for third-party testing, and focus on three things—actual coverage in concrete-wall environments, stability under concurrent multi-device load, and uplink speed. In early user feedback, statements like "the camera at the door no longer drops offline" are far more credible than "it's several times faster."
Pure tinkerers: what you should really watch is not this machine, but whether that RPU will be integrated into mainstream SoCs. DPD is standard in base stations; the reason it has never made it into home routers is that the cost cannot be pushed down. If the next generation of mainstream Wi-Fi chips makes it standard, even hundred-yuan devices can get it—that is what this means for ordinary people.
Five things to clarify before placing an order
- What the uplink speed actually is. All the promotional numbers right now are downlink; the official side does not mention uplink at all.
- Under 256MB RAM, whether the NAT session table and "more than 100 devices" can hold up at the same time. The official materials only say it can connect, without saying what state it is in after connection.
- How to buy in mainland China, who bears tariffs, and where to send it back if it breaks. The shipping regions during the crowdfunding stage include mainland China; the current direct-shipping policy on the official website needs to be confirmed by yourself.
- Whether third-party firmware can be flashed. The official materials do not mention it. MT7981 has good third-party firmware support on other machines, but this unit adds a self-developed chip, so it most likely will not be opened up.
- Whether two units can form a Mesh only with their own brand or can be mixed. The chassis has a Mesh button, but the official materials do not say what it is compatible with.
In one sentence
The real value of this is that base-station-grade digital RF processing has been packed into a consumer-grade chip for the first time; the direction is right. But in this version of the product, both price and configuration are still at the stage of "selling to the first batch of people willing to pay a premium"—domestic NAS enthusiasts and 2.5G users do not need to look at this generation; those who want to buy it for single-device coverage in large homes should wait for third-party real-world tests; those who only care about the technology direction should remember the three letters DPD, which will most likely appear in much cheaper machines within two to three years.
Sources for this article: WavKong official product page (accessed November 1, 2026; price 520 USD and struck-through original price 703 USD, platform MTK7981 AX3000, 128MB flash / 256MB RAM, 2.4G and 5G each with 2×2 antennas, 1 Gigabit WAN + 3 Gigabit LAN, Wi-Fi 6 dual-band 160MHz and 1024-QAM, WPA3, 12V/4A, 281×166×79 mm and 1.32 kg, router / bridge / AP+client modes, AWS North America cloud, Wi-Fi 6-related FAQ; prices are subject to the current official website prices); WavKong October 30, 2026 launch press release via PR Newswire (commercial launch date, 15,000 sq ft combined indoor and outdoor, tests in single-story wood-frame and three-story wood-frame houses, garden test point 123 Mbps vs. 4.2 Mbps, EVM from about −29.3 dB to −41.2 dB, Kickstarter launched April 22, 2026 / 1,074 backers / 3,272,087 HKD / delivery completed September 29); Kickstarter project page FAQ (self-developed RPU ASIC has completed tape-out and validation; shipping regions include mainland China); IEEE 802.11ax and LitePoint public technical documents (the EVM requirement for 1024-QAM is no higher than −35 dB); Xiaomi official website AX3000T product page (dual-core 1.3GHz, 12nm, 256MB RAM, officially claimed to stably connect 128 devices, laboratory conditions); JD.com November 1, 2026 page (AX3000T regular selling price 129 yuan; subject to the current official price); SBCwiki and MediaTek public materials (MT7981 is the Filogic 820 entry-level Wi-Fi 6 SoC, dual-core Cortex-A53 1.3GHz, 12nm, native support for 2.5GbE HSGMII, used in the OpenWrt One reference board). The parameters, prices, and test figures in this article are all relayed from the above sources, with no currency or unit conversion; judgments about uplink, wall differences, memory and concurrency, and third-party firmware support are personal analysis and have not been tested on this model.
