Hacker News new | ask | show | jobs
by andrewstuart 1208 days ago
This is the reason why you can't get a Raspberry Pi computer.

Eben Upton, (Raspberry Pi Founder) sold out his his base.

When the chip shortage hit, Upton made the decision to prioritise supply to industrial buyers over retail/hobbyists/education.

His rationale was that companies that had bills to pay and salaries to pay should get priority because people's livelihoods depended on getting those Raspberry Pi's. Sounds good. But years down the track, nothing changed. What should have happened is that Upton should have given the industrial buyers a six months warning to find other ways to build their products instead of Raspberry Pi. That never happened. Instead, Raspberry Pi simply turned into a dedicated industrial computing supplier. Businesses have the means and the expertise to redesign their products around different computing technology - they did not need a blanket commitment forever from Raspberry Pi to support their technology directions.

The outcome of this decision has been that Raspberry Pi and Eben Upton have sold out the community that got Raspberry pi to where it is.

He sold out the kids and the schools and sold out the hobbyists.

I think that in the Covid time, if anything, he should have prioritised schools and children. When they were sitting at home in lockdown they could have been learning computing with a Raspberry Pi.

So its now years down the track and nothing has changed. You can't get a Raspberry Pi because they are sold to industrial buyers like Vodafone.

Raspberry Pi's loyal community and especially the kids and the schools deserved better than being sold out.

There's lots of alternatives to Raspberry Pi - give them your loyalty and your money.... here's some:

https://www.youtube.com/watch?v=k8clrUclPIs

https://www.youtube.com/watch?v=UGcWLgAk5kA

9 comments

Raspberry Pi is a corporate endeavor as well. They also have employees, and a payroll to maintain.

Seems a bit harsh to blame them for selling their products to the customers who are most able to pay for them, which enables them to stay in business. (Whatever quantity Vodafone is buying is DEFINITELY more than any hobbyist is able to pay.)

https://www.bbc.co.uk/news/technology-17190918

You're very wrong, that was not what it was for.

Dr Upton says that will help the project grow much more quickly than previously thought.

"We didn't realise how successful this was going to be," he said.

"This means we can scale to volume. Now we can concentrate on teaching people to program."

What a lie that turned out to be!

And yet the annals of history are littered with open source education efforts that failed, because they weren’t able to figure out a revenue model that was sustainable to their creators.

Maybe this thread is just a lightning rod for a bunch of OSHW anger but there’s compounding good that comes from a huge customer like Vodafone buying a shitload of Pis.

The major thing I can think of is higher volume for the Pi foundation, which means more negotiating power with their chip suppliers or price breaks at a certain ordering volume.

Might mean you don’t get any Pis now.

Also might mean they drop the per unit cost by $5-10 in a year.

Not that I expect any of these facts to make you feel any less angry or betrayed or whatever, but they’re net goods for the Pi Foundation, and the OSHW community if they do indeed come to pass.

> The major thing I can think of is higher volume for the Pi foundation, which means more negotiating power with their chip suppliers or price breaks at a certain ordering volume.

Normally you'd be right. But apparently (the unsubstantiated rumor is that) Broadcom is pissed that RPi built a for profit company on top of the IP licenses Broadcom thought they were donating to a nonprofit and is stone walling RPi wrt getting larger orders than previously negotiated or working with them on an RPi5 and that's the major issue with the continuing stocking issues.

That wouldn’t surprise from Broadcom, they’re some petty MFers to work with in my experience lol.

Seems like they’re cutting off their nose to spite their faces at this kind of volume, though.

> Raspberry Pi is a corporate endeavor as well. They also have employees, and a payroll to maintain.

They would have sold out either way. What does payroll have to do with who they prioritize to sell to?

It's easier and less risky to sell in-bulk to another business on a contract than it does to sell retail
Another good point. Not to mention probably a more profitable end venture for the Pi Foundation. Fewer pounds of flesh doled out to retailers and middlemen.
> Whatever quantity Vodafone is buying is DEFINITELY more than any hobbyist is able to pay.

You're arguing for the parent comment with this.

The "foundation" was intentioned as for hobbyists and people who wanted to learn while being affordable that's how it was sold to the general public. Chasing a higher payer will always lock out "affordable".

An analogy would be healthcare. Capitalism and affordable+for-many are at direct odds.

> Chasing a higher payer will always lock out "affordable".

This reveals a misunderstanding of the economics of both chipset vendors and electronics manufacturing.

The more Pis that the foundation can make at one time, the cheaper their unit costs can be, by encouraging competition from suppliers and getting price breaks from chipset vendors.

Large customers like Vodafone buying 10k or 100k is a huge advancement towards that goal.

In the short term? Yeah, maybe some shortages of SBCs.

In the long term, they’ll probably be able to build more Pis at higher quantities and cheaper unit prices.

> In the long term, they’ll probably be able to build more Pis at higher quantities and cheaper unit prices.

My retort is - then you misunderstand the gap between Android and iPhone. MediaTek+Qualcomm vs Apple [0]

It's too simplistic to think just because you produce more you'll make more money. There's an absolute maximum in the pricing-volume curves. Even if you have to pay more for the chips, if you can simply charge more you'll make money hand over fist. That's what this thread is saying that the RPi foundation "sold out".

The last sentence is incredibly hard to achieve, but if your brand is a household name like iPhone or R.Pi then you can start jacking up the price and as long as you have correctly identified the price insensitive (or less sensitive anyway) you price to the max of the particular group.

[0] https://www.counterpointresearch.com/global-smartphone-ap-ma...

Not every brand makes the decision to raise prices in the face of lowering backend costs, however.

I don't have enough info on historical pricing of RPis to say, but it certainly seems to benefit their stated mission of education (and unstated one of industrial SBC supplier) to have lower per-unit pricing at retail or wholesale.

Though I get the feeling I may be arguing in the face of a bunch of folks who just want to mob the Foundation for taking a course they don't personally agree with.

Many capitalist countries have affordable healthcare.
You can bitch about RPIs or you can go support something else.

Lots (9000+) of Beaglebone Blacks over here -> https://www.mouser.com/ProductDetail/BeagleBoard-by-Seeed-St...

TI documentation is stellar (except the graphics--which nobody documents) unlike the RPi or any of it's clones. The chips are all available at volumes less than millions. You don't have to sign NDAs in order to get stuff.

The Linux is standard Debian with everything upstreamed. You have lots more pins. You have two real-time units. It doesn't need special overpowered power supplies or heatsinks--USB 2.0 500mA works just fine even without a heatsink. And you have onboard eMMC so you don't have to wonder whether the craptastic uSD card that came with your RPi is causing those strange reboots.

And the folks doing it all are woefully understaffed and undersupported.

So, everybody can sit and complain about lack of RPi's. Or they can go support something that is actually open source, available and delivering.

Yes, they had their supply issues like everybody else. But TI is happy to actually move chips in ways that Broadcom simply isn't.

> You can bitch about RPIs or you can go support something else.

I will and thanks for your comment... I'm pretty much Debian and Debian derivatives only so you convinced me.

I've got eight Pi I think. I started with the "Model 1" (which wasn't called that back then), on which I was running RasPBX (Raspbian with FreePBX+Asterisk) hooked to a SIP trunk to power four Cisco VoIP phone at my wife's little SME. Worked flawlessly for years. My best one I'd say is in the vintage arcade cab, with a Pi2JAMMA hat bigger than the Pi (!). I've got three towered on my desk, two of them operational 24/7.

So early adopter and user as it gets, not the kind to let the Pi not powered up... But I think this headline and the troubles getting one convinced me and I'm done with that brand.

I'll order BeagleBoards now and run Debian on them.

Sorry I came across a bit salty. However, I've been banging on about this for ages.

The RPi foundation has been subsidizing board cost since practically the beginning. This is nice in that it puts boards in everybody's hands.

This is bad in that it sucks the oxygen out of the space since anybody who doesn't have a benefactor willing to throw cash to subsidize the cost gets pushed to the side. Then, when the money finally dries up (and it will always dry up), the space immediately becomes a barren wasteland because everybody else got driven off.

It's doubly bad since mere mortals couldn't actually get their hands on the SoC chip at the heart of the boards. So, nobody else could take over when the RPi foundation decided to bail out.

Which is where we are now.

It's not perfect. There was (not sure if they fixed in subsequent hardware revisions) an issue with the ethernet not working about 50% of the time, and this required cutting the physical power to the device to fix. https://community.element14.com/products/devtools/single-boa...

I found this out the hard way after building an IOT vehicle logger using these boards, had to work in a hardware self-reset to rework the board.

Was that specific to the BBB industrial? That's a really unfortunate bug.

I agree, about the "nothing is perfect", though. The lesson when you do stuff in embedded really is "Trust nothing."

The cheap subsidized Broadcom part is a habit that's hard to break. I've stopped suggesting iMX or Sitara to people because they just can't get over what the real price of these things are. And it's not hundreds of dollars.
I remember when the pi came out. It was supposedly targeted towards third world countries that couldn't afford computer hardware. That changed very rapidly as soon as they saw how much they could profit off of hobbyists. This shift towards the corporate customer is just more of the same.
I've literally just bought a 2 year old 10nm Tiger Lake Intel i5 16GB, 512GB_ssd, Lenovo Slim 7i notebook, on Ebay, for 400 USD. Thanks to M1/2 these devices are going for cheap: 500 dollars in general for like-new (mine had a few scratches), and they perform 5x better (at least) than an RPi 4, have at least 2x the RAM, have an actual SSD, keyboard, monitor, backup battery, power supply. Mine even does AVX512. The SBC space has become very poor value compared with recent x86 devices cast aside because of the (admittedly amazing) macbooks. Admittedly RPi has a good software support story but the x86 devices are gold standard here and they're crushing the ARM SBCs right now on value. You can get a 150 USD full-fledged notebook with a Celeron that is 50-100% faster than an RPI and includes 64GB of Emmc, screen, keyboard, power supply, and impeccable software support, in the form of Asus L210, and others.I still have an RPI4, I like it, but the Upton interview on Explaining Computers in December really put me off because he sounded like a banker or corporate consultant. The RPi ethos is gone. And so is the value.
That's not the same thing at all. 400 USD is, well, way more expensive than a Pi.

Then there's the room it takes / the formfactor. I've got three Pi stacked on my desk (and 2 out of the 3 are powered at nearly all times). I've got one in my arcade cab, with a Pi2JAMMA adapter. The Notebook wouldn't fit or would look clunky. Also, well, how do you put a hat on a Notebook? How do I put it in my cab and have the JAMMA connector power it?

You're comparing apples and oranges (not that your notebook doesn't look good).

I'm not saying what they're doing is fine but people looking for a Pi aren't looking notebook.

I think this is an uninformed knee-jerk reaction. Eben has been very public about the supply chain issues, the prioritization of small companies that rely on shipping stock to keep afloat and how mainstream stock will be replenished (it is already possible to buy Pis again—I have an RSS feed for rpilocator and the number of updates increases daily).

Vodafone is doing the Vodafone thing and pushing out a _prototype_ that just happens to have a Pi managing a $400 RF board. I worked there for 11 years and have seen this thing time and again (I was a product manager for broadband and CPEs, and remember the Linksys WRT54g-3G, a bunch of Sagem and Huawei mini base stations, etc.). You are completely misinterpreting the reach of such a prototype, no matter what the marketroids spun around it.

It’s Mobile World Congress week, so the telco distortion field is at its yearly zenith.

This looks interesting as an alternative:

https://www.sifive.com/boards/hifive-pro-p550

> "This is a premium software development system ideal for developer desktop machines and rack-based build/test/deploy servers for RISC-V software development. RISC-V has no limits."

I don’t seem to be as phased or disappointed by the Rpi shortage as everyone else.

There are so many alternatives that are often cheaper and more powerful. Here’s a recent summary of the alternatives:

https://youtu.be/uJvCVw1yONQ

I think you are wildly overestimating the benefits of giving children raspberry pis.
An uplifting quote from: https://www.cam.ac.uk/stories/raspberrypi

"Raspberry Pi wasn't invented to boost shareholder value or turn its founders into billionaires: it was created as a charity to increase the number and calibre of students applying to study computer science at Cambridge, to give young people access to programmable hardware at a low price, and to equip a new generation"

Whereas Raspberry Pi today:

"We're selling to industrial buyers."

That was 10 years ago; those then-students are now engineers. This is what success at equipping a new generation looks like - the transition of needing tools to do stuff with[0] and not just play with and learn on.

[0]"Industry"

I mean, most educational tools don't just decide to give up on students when the first cadre age out.
There are so many other SBCs as well. Arguably the arduino products are more useful to kids anyway. Much easier to use, you just plug them in to the computer you are already familiar with and can start running code on them. No need to set up SSH or a second monitor/kb
But no easy wifi
ESP32 has wifi and bluetooth and costs hardly anything.
Perhaps, but the Raspberry Pi Foundation was founded on the premise of putting tiny computers in the hands of children and hobbyists. So whatever good may come of selling RPis to Vodafone, it is not the good that Raspberry Pi was founded to provide.

This is a GPUs-in-the-hands-of-miners-not-gamers situation, but I think starving the educational community is a bit more venal than starving the gaming community.

this makes me nervous to design anything substantial around an RP2040

with a Raspberry Pi at least you could port your application to any board that runs Linux just by moving the code over

but if your project depends on the PIO subsystem in any way, you're locked in to buying their chips forever; you can't just take your code to another platform

RP2040 doesn't have any Broadcom IP blocks, so it's probably a bit safer. About what I'd expect out of any other MCU that I didn't have a multi year buy signed in the blood of newborns for.
At least the RP2040 chip is appearing in boards from other manufacturers, like Arduino
Why does it have to be Rpi for those kind of project, I am wondering? What stops me from buying a NIC or even a bare-board PC for the project?
A NIC? Or a NUC? I am not sure exactly how Vodafone have packed the boards in their little black box, but presumably you could DIY something similar, though I guess it helps to be a carrier for making th SIM? eSIM? work. It is sort of odd they didn't build it around the Compute Module though, I thought it was really intended for these commercial customers.