Prediction: The probability of YoLink releasing a “local” hub by January 1, 2025, is 10%

Hardware
1

Prediction: The probability of YoLink releasing a “local” hub by January 1, 2025, is 10%.
The rationale for this prediction has 5 parts.

  1. YoLink products are now available at Home Depot, Lowes and Walmart. These are high volume retail distribution channels with a broad customer base. The customer base is not sufficiently sophisticated to see the advantage of local vs cloud architecture. YoLink is selling into the broad retail market, not the hobbyist market.
  2. YoLink is no longer supporting Discord, Reddit, and Home Assistant forums. YoLink’s forum facing presence, Eric, has disappeared and has not been replaced. The population that inhabits these forums is no longer YoLink’s marketing target.
  3. The return on investment for developing a local hub is unclear. The demand for a local hub is limited to a small, albeit sophisticated, market. The development effort seems to me, from the outside, to be more than a few months to less than a year.
  4. Providing a local hub to the retail market would places YoLink in the position of supporting two different architectures. Supporting a local hub will increase expenses.
  5. YoLink has not released local hub specifications or a delivery date. The specifications, development, test, fabrication, test, documentation cycle for a commercial product is typically lengthy. Yes, YoLink said they were going to develop a local hub, but ….
    There are individuals on this forum that are anxiously awaiting a local hub. My prediction is that this is not going to happen.

I am impressed by YoLink products, so I did “due diligence” on YoSmart and YoLink. The results were rather opaque. The address of YoLink, 25172 Arctic Ocean Drive Ste 106, Lake Forest, California 92630 looks like a mail drop. A google search on the address returned several companies with the same address. When I attempted to pull a D&B credit report on YoSmart and YoLink, D&B did not have records on either company. The YoLink web site contains the description “YoLink, a division of YoSmart, a US corporation based in Irvine, California, ….”

My YoLink investment is small, a few hundred dollars and it works so I am going to keep using it.

I hope this is interesting. I am going to cross-post this on Home Automation, Reddit and Discord.

2 Likes
2

From hub3 page:
As of April 2024, the YoLink Hub 3 does not support local integration. It is planned to be completed by end of year along with matter compatibility and once completed, will be through a software update by the Hub 3.

3

I’ve been checking every couple of months and wanted to add a couple observations.

In reply to 3 & 4 - by adding support for matter, they gain the attention of a growing audience of consumers who value interoperability. The HomeAssistant integration would become redundant at this point, so they could likely drop support for it (whatever minimal support exists).

Regarding hub 3 - there is a glaring problem which remains unaccounted for, if my understanding of the current architecture is correct.

Following is my personal speculation:

My limited understanding is the existing protocol uses the hubs as simple border routers, meaning they do not have knowledge of the message contents or a way to check validity. This seems like a reasonable design, as it’s more secure if the private keys are not distributed into the field at all. Thus it is not possible to replay messages or snoop the contents, or issue a fake firmware, among other concerns which could spell trouble for the whole product line if messed up. It also has the advantage of expanding reliability and coverage, since any hub within range can forward messages to the cloud.

With matter or local control, the whole situation changes. There are tradeoffs, but ideally, they want to keep the existing level of coverage and not pair the sensors to a specific hub. Thus you could have multiple hubs forwarding messages to HomeAssistant, and the messages can be deduplicated easily with a sequence number.

Option A - The hidden complexity is whether to implement matter at the sensor level - in which case end to end encryption is possible with HomeAssistant (the hub needs no knowledge of the message contents). This is a really simple way to handle key distribution, since it effectively operates the same as it would in the cloud. It’s a straight forward implementation where HomeAssistant actually pairs directly with the sensors.

Option B - If they want to take a shortcut then they can try to implement matter at the hub only, with minimal changes to sensor firmware. But the time saving is replaced with the complexity of distributing keys to the hubs in a secure manner, and the associated pairing process which currently exists only for devices with “D2D support”. By allowing the hub to open messages, they are risking the possibility of a huge breach where a method is found to extract or replace a key on the hub, or any other security flaw, thus compromising any/all customers in the general case. It’s a lot of hoops to jump through and a lot of possible flaws or attack vectors to account for since each hub is effectively a target. Whereas previously the consequences of a compromised hub are less severe, with less attack surface area in the first place.

Option C - they can try to retrofit a mesh network, or just forego the coverage entirely if using matter and rely on only one hub / some compromise of the above.

Option D - customer managed PKI - the could essentially give customers a docker image of the YoLink cloud infrastructure, assuming licensing and intellectual property concerns are accounted for, and “release” the devices from the cloud, effectively orphaning them. This is the option which would serve only HomeAssistant type users, and makes least business sense.

Politics of the decision to add local support are (hopefully) taking a back seat to the security concerns, and the ROI has to take the possibility of bad press/bricked devices/threat actors into account. If Matter and mesh support existed from the beginning then this wouldn’t be a hard ask.

The upside of a proper and secure Matter implementation is that going forward, the YoLink app does not need to continue reinventing the wheel regarding new features which have overlap in other platforms. Essentially it would be like having “10% off” on their cloud bill, if data exchange for Matter enabled devices is kept local only. It’s a write once / reuse situation if they get it right on the first try. But it is also like changing the tires on a moving car and then having to send it in for crash certification and safety ratings due to the size and scope of the change. Not without risk.