It’s been a few weeks since I had a few interactions with the Intel Corporation, so I thought I’d write up a summary of what I think they’re screwing up from the small-developer / indie-developer perspective, in case whoever they choose as their next CEO is open to actual, on-the-ground perspectives.
The start of the story goes something like this: I had an idea about building a piece of electronics using their Thunderbolt 3 connectivity chips. It wasn’t such a huge idea, let’s say, but it would’ve been interesting to me, and it provided an excuse to learn more about the technology.
This is as good a reason as any for most engineers to start digging into something.
I understand, of course, that companies may not have the resources to engage with every person that comes knocking. But datasheets and reference manuals are sunk costs anyways, it costs nothing to give people access to a downloadable PDF. It’s not like they would have to find someone in the mail room to print, pack, and ship something to me.
Now, Intel is currently the sole supplier of Thunderbolt-certified chips and chipsets. In order to build anything Thunderbolt related, you have to gain access to their Thunderbolt Development web portal.
No problem, I thought, naïve as I was.
Instead of making their datasheets available openly for people to study and begin to think about, the Thunderbolt datasheets are hidden behind an access-controlled website and require an NDA and developer’s agreement to access. You don’t just sign up, get a confirmation email, and then click right through to the information you need.
I mention this to a developer friend of mine, and he just shakes his head and laughs.
So I sign up on the web portal and get an email from the person nominally in charge of it that he’s currently attending Computex 2018 in Taipei. Just a simple “OoO for Computex”, no additional information about who else could handle this.
It literally looks like this:
I wait 9 days to hear something back from them, before finally emailing the person in charge again. It wasn’t time critical for me to get a response, but I could only imagine someone actually trying to do business having to deal with a 2 – 3 week response time?
This time, I get an autoresponder. The person in charge had quit, and their last day was the day before. Something tells me the morale in this department might be suffering?
Then, I finally get a form-letter response from the the Thunderbolt management, and it royally pisses me off.
Here’s what they say:
Hi Max,
Thank you for your interest in developing an eGFX enclosure using Thunderbolt™.
The eGFX market is still developing with limited volume available for all vendors. And as we already have a number of eGFX devices under development, we are not accepting applications for eGFX enclosure designs with new electronics at this time.
However, if you still want to participate in the market – and we hope you are – white-box solutions and turn-key eGFX electronics kits are available through a few of our device manufacturers. And although you will need to submit these devices with your brand and/or enclosure for re-certification, we believe these options allow you the opportunity to greatly reduce your development effort and cost. In addition, the re-certification process is significantly simpler than qualifying and certifying a new design.
Contacts at the Thunderbolt eGFX vendors who are open to providing you white-box solutions or electronics kits for your eGFX enclosure are listed below:
• Sonnet (white box solution): [redacted]
• Akitio (white box solution): [redacted]
• TUL (turn-key eGFX electronics): [redacted]Once you have confirmed your approach and solution, please contact us with your updated application and we will help guide you through the process. Note that you will need to sign the Thunderbolt Trademark Licensing Agreement to proceed at that time. After licensing process is completed we will be able to grant you access to the Thunderbolt Developers Area.
We look forward to hearing from you.
Thank you,
Thunderbolt Admin
So let’s get this straight. You have to “apply” to use something a company ostensibly wants to sell. They have to “bless” your idea? And you can’t see anything until you have signed a bunch of licensing agreements and other red tape.
Intel proves in writing that it’s got no clue how to engage with small third parties, which is the long tail of development that is happening these days. No one wants to deal with this just to gain access to information that may or may not be useful.
Besides, I wasn’t building exactly an eGFX device, I just told Intel this because it’s the closest thing. Too bad that they think of the world in such black and white terms.
Here’s what I write back (admittedly pretty angry):
Hi Intel,
This email response to my application to the Thunderbolt Developer Program basically just makes me feel angry. Couple that with the fact that it took 11 days to even get this response.
This is the second instance in as many years that I feel like Intel let its developers down, the first being the introduction and abrupt cancellation of the entire Curie / Edison lines.
You guys realize this is why people often form a bad opinion about working with you, right?
Intel can’t have it both ways:
Intel cannot say they want tons of people to use and adopt the Thunderbolt standard and then go on to tell people who want to experiment with their chips that Intel has the right to pick winners and losers, and that my best option is to work with a white-label ODM.
You want to control the ecosystem that badly? To tell engineers that they should focus on product marketing and packaging instead of differentiation or innovation? You want _less_ competition in the peripherals space?
Maybe I don’t want to take the easy road (engineers often don’t), maybe I don’t want to build _exactly_ what these ODMs are building, and maybe I don’t want to have my revenue floor be set by a go-between supplier. Maybe I’m a maker who wants to start experimenting with Intel’s chips in raw form or based on a reference design that you publish. I certainly don’t have a problem assembling my own PCBs or putting together a BOM. Maybe I would fail, but in the process learn a bit about the Thunderbolt controller chips that I could use in a different product sometime later.
Why are you afraid to let people play with the chips you spend so much money developing? If the devices that come out of this low-cost tinkering are garbage, you can stop them at certification anyways.
Contrast this with the way ARM does business. Everyone can license and produce products based on core technology. The designs and reference manuals are easy to acquire. ARM doesn’t pick winners or losers. And they don’t limit the upside. End of story.
Intel is toast if it doesn’t see that it’s one-by-one making engineers not want to do business with it, from the smallest fish upwards. You’re a single-source, and you avoid open distribution channels. That ALREADY scares enough people away, but when you even turn away the ones who come to you anyways to see what the hubbub is about… Wow. Just, wow.
Have a nice day,
Max Vilimpoc
To their credit, Intel actually follows up with me and schedules a teleconference where I get to talk to a product group manager about how badly they seem to be managing the proliferation of their very-expensive-to-develop technology.
The manager informs me that Intel is in fact working on getting the Thunderbolt technology to be a more open standard, so that more suppliers could start producing compatible transceivers. This is good news. But the wheels turn slowly, so there isn’t yet a deadline or schedule informing when this will happen.
I ask him if there’s any way he could just grant me access to the datasheets in the website. He equivocates and never grants access. The call ends without much sense that things would change.
I get the impression that the changes needed will happen on a geologic time scale. So I completely stop thinking about working on anything to do with Thunderbolt.
And so, one more engineer decides it’s a better use of his time to focus on other things.
Intel can’t afford to keep interested people at arm’s length (pun intended), even if it’s just one or two engineers at a time. The fact that they keep doing so does not bode well for their future prospects.
Now, this is only one sample point from possibly thousands of interactions that the company has every day with developers, but it seems to me that the long-term sustainability of any platform or technology usually starts with getting people to use the platform at all.
So why would Intel be interested in turning away anyone who shows the remotest interest in their technology?
Or, as Ballmer would say, more succinctly: “Developers, developers, developers.“