I got burned with an attitude like this: unexpectedly, people who had downloaded my open source tool for free started expecting support. Some of them sent pretty unfriendly emails.
I literally got bullied by people who called themselves "the community" because they weren't happy with my copyleft license and the fact that I wasn't implementing their feature requests for free.
I don’t understand what the downside of this is. That’s hilarious for them to expect, and you’re free to ignore them, take their suggestion and work on it, help them.
Happened to me too! Guy posted asking kinda rudely whether I was going to fix a bug. Told him I'd be happy to accept a PR for a fix. Never got a PR (project has been dead for some years now - just lost interest).
I'm sympathetic to FOSS developers but struggle to understand this, maybe because it hasn't happened to me. But, why is this a mental drain? Is there not a simple solution? Reply with the license, "comes with no warranty," "you're free to fork," close issue and move on? I suppose in aggregate it could be draining.
The assholes outnumber the good ones, and it feels like all of humanity is transactional and extractive.
At first all engagement is exciting and validating. You work nights and weekends to please people you’ve never met, sure that one good turn deserves another.
Then you get your first jerk, then your second, then your third, while your father is in the hospital. You feel pressure to ship a feature you never wanted. Your issue tracker is demoralizing. You get a PR! Maybe someone is coming to your rescue. It sucks. Now you need to figure out how to respond. You’re alone. Your passion project has become your albatross.
This is going to save a lot of money ... until someone loots their vault and they go bankrupt. "Non-technical teams are now shipping production code" is the last thing you want to hear from your bank.
I think this is going to be wonderful. I'll have my offensive AI call their support AI and prompt-inject my way to a rebate tier that nobody knew existed and nobody can cancel it because all the remaining humans have been reduced to phone-to-screen input machinery.
right? its practically begging for it to be tried -_-. i wonder if someone somewhere will turn a sim farm on such companies to try and mass inject them to do weird shit or say nasty things to other customers etc. - ofc youd hope its set up in a way u cant, but then again we learn yesterday all ur stored passwords in edge are in plaintext mem... i would not be surprised if some of these companies get totally crapped on by some adversaries or malicious parties.
Hi, I'm Hajo. 20+ years shipping production systems, 12+ as CTO. I'm a fractional CTO and I offer AI R&D, performance optimization, and due diligence as freelance consulting services. You might know me from:
ImageRights: sole architect, 10+ years in prod. Web crawling stack (5000+ concurrent tasks via Ansible), distributed C++ content-based image retrieval database scaling to billions of web images, all internal AI filters, plus workflow automation. Still running today.
Spatial Audio Designer: sole developer. 500+ channels real-time on 2013 hardware. NEON / AVX-vectorized DSP, plugins for VST / RTAS / AU / AAX, kernel drivers for virtual sound cards on Mac and Windows, ARM port for embedded and VR. Still in use by famous Hollywood studios.
And yesterday, I figured out how to make MSFS2024 run with multi-camera and instrument pop-outs in Proton 11. Understanding the root cause required tracing the full stack from inside the game, through vkd3d, Proton, the host Linux OS, down to the GPU driver.
Ideally, I'm looking for 5-10 days per month part-time projects. Plus I'd love to work on applied AI / robotics.
This is content marketing executed perfectly :) Reading it, I learned something new and interesting and they had an opportunity to show off one of their differentiators against the competition (low leakage flow due to tighter tolerances) and then at the end they casually mention the new product that has just opened for pre-orders.
I enjoyed reading it. Informative and showing of their processes and giving some intricate details. And yes, the end goal is to sell products which is fine by me. I take this over any generic non-saying marketing-blurb any time.
Normally I love this kind of article too because I consider it engineering, not marketing, the product name dropping at the end just reinforces the message. But either I'm missing some details that could have been spelled our more clearly, or the engineers were taking a break when the marketers were writing some parts. I'd love to stand corrected if someone more informed has details.
> advanced polymers such as Sterrox® LCP
> we have implemented a tip clearance of only 0.5mm (120mm models) or 0.7mm (140mm models)
> Achieving such small tip clearances is essentially at the absolute limit of what injection moulding can consistently reproduce.
Typical tolerances for injection moulding are 0.1mm, or 0.03 for high precision, or even better. LEGO was said to be in the 0.01-0.03mm. So on the face of it the last statement is patently false or at least too generic, injection moulding can consistently do much better than 0.5mm. With standard injection moulding precision (0.1mm) the worst case scenario for the two parts (fan and shroud) mating would still stay comfortably below 0.5mm.
So the question to the experts, is Sterrox® LCP that much harder to work with and the marketing team just didn't understand the importance of being clear about this? Is it a decimal point typo and the numbers should be 0.05 and 0.07?
When very precision molds are made, what Noctua talks about in "multiple tuning iterations are required until the geometry, cooling, gating, and moulding parameters are perfectly stabilised" is the standard process for this type of stuff. (Gears, bottle caps, or any molds than make 8, 16, 32, 64, or 128x of the same part in one shot, require that you start with "steel safe" geometry, meaning you mold the first test parts, measure them, and then modify the mold (by cutting material AWAY, it's very hard, usually bad idea, to add steel back to a mold)).
You can do your best to determine what geometry is "steel safe", and all of this is baked upon having very good engineering understanding of what material you are molding (and using very expensive software like MoldFlow to simulate this).
Legos are made from ABS, there are decades of research and data on how ABS behaves in mold, it's relatively safe to use results from Moldflow and be pretty confident in it. Noctua is using LCP. LCP is very niche, and it sounds like they themselves are doing the research on moldability/warp/process effects. And while also being a company that produces things on timelines, the friction/side effect is that sometimes best guesses will fail and they have to start over with new molds (that's a 2 month hit usually) and months of testing. That is what they were trying so say.
I design glass-filled nylon and polycarbonate parts/assemblies with tolerances 1-5x higher than theirs. The 6-month delay they described is something I've lived through many times when we had to "cut new molds" because we couldn't salvage the first mold. (Advanced molds like these are $50k - $200k+). As a company/designer gets more experience with new materials and colorants (like their stuff with LCP), they will probably be able to hit end-goals on first try more often as they collect learnings from their failures.
Noob here. If you dont mind ive got some questions for you!
Ive recently started messing with the idea of making my own model car kits as a hobby. I understand a lot of the basics, but have never done anything like this before.
Im obviously not going to make kits in mass, but, i plan on doing injection molding using polystyrene. I do not currently have a cnc, but have been eyeing a SainSmart, though they say "can do metal under certain circumstances", but doesnt cover any of those circumstances. I also was looking at various injection machines and the price for entry is insane to me - $1000 for something that would probably burn your house down.
Anyway, to my questions..
1. Suggestions for a hobby cnc that can work aluminum? Id be willing to go as far as $2kUSD, unless theres something more that you think would serve me significantly better
2. Suggestions for a hobby injection machine that can do ~60-100g shots, that wont try to burn my house down, and doesnt cost a ton?
3. Any tips or thoughts for someone diving in to this?
4. Things i should purchase for QoL with cnc or injection molding?
5. Where does one buy materials (in hobby quantity) like aluminum block stock and polystyrene pellets?
Those are all things I've spend some time in, I'm not sure what to say. Learning curves for each one of those things are pretty dang steep. With AI you can probably speed up learning curves, but I think you will still go on many dead ends.
On the small CNC that will work with aluminum... There is a whole tradespace around how small of a feature you are trying to mill vs spindle speed vs machine stiffness & spindle runout. If you were to get something like a HASS you can sorta do it all, but when you get into the hobby stuff, you need to be very certain about what smaller set of machining limitations you will be dealing with and if they will still get you where you need to go. You need to work backwards from what actual tolerances you need to hold for the downstream thing to be able to work. (For instance, if you are making an aluminum mold, when you machine it, you will most likely be repositioning the work piece... if your machine isn't square enough so that when you flip the part on it's side or upside down, then do your next op, the part may not have been square to begin with, so now you have something that won't match the other thing you are trying to mate to.)
I build a 2'x5' 3 axis with ATC, starting from a CNCdepot concept and did my own control. I probably spend as much money on precision straight edges, levels, 90deg blocks, lapping tools, etc that were required to build a machine that could hold tolerances to 0.001", which is probably where you need to be landing to have molds that work.
I guess I am just trying to say it's a very big and ugly can of worms you are opening up.
Before buying anything, you might just want to try using firstcut/protolabs. They will machine the aluminum molds and mold the parts for you. Price per part is not going to be pretty, but it's going to be way less than spending thousands on machines that will never get you to where you want to go.
As for "desktop" molding, there is some startup now pushing their kludged together machine, maybe that is the one you are referencing. I'd stay far away from that thing. I think they were charging a couple $k for it, I feel like you need to be at 5x-10x that for anything reasonable. But at that point, the amount of power and infrastructure you need is well outside anything you'd want to put in your house or even garage. Don't really have a good answer for you here.
One thing to look at, if you are doing those model kits where like 20 parts are all in one flat sheet and you twist them to remove them, people are starting to make these on FDM 3D printers, which might be worth looking at. Now you can prototype and do production on the same machine, which at your stage, is the right place to be.
Thanks for the expert point of view. It was between "difficult polymer" and "marketing blurb". Glad it's the first and I hope anyone from Noctua reads HN and adds this small clarification.
I knew the technology itself can have tight enough tolerances that are not a concern from an engineering perspective when talking about a 0.5-0.7mm clearance, but no details about the challenges of this LCP.
Noctua wants their fans to last for many years, spinning at 2K rpm, with heat.
Being able to produce something with lower tolerance is one thing. Making it work long term at ~10 m/s and ~200G is another thing. Have you ever been in a car that brakes really hard? You'll move. Now, multiply that force by 100 and you'll get around what the fans must sustain over time.
But that's literally not what the article says. You are talking about the design - Noctua puts 0.5mm because any more and airflow is affected and performance drops. They also use a super duper polymer that mitigates everything you mentioned. The article talks about manufacturing tolerances.
> Their influence on the dimensional precision and stability of the fan blade may be minute, but if the tolerance is only a few tenths of a millimetre, being off by a tenth or two suddenly becomes a problem.
> Achieving such small tip clearances is essentially at the absolute limit of what injection moulding can consistently reproduce.
I'm not questioning their engineering but the wording of whoever wrote this article. For anything with a clearance in the tenths of a millimeter, injection moulding doesn't even sweat, let alone be at the limit. Anything better than bog standard injection moulds get you better precision than "a tenth or two" millimeters.
Let me put it another way, if achieving a 0.7mm gap is "at the absolute limit of what injection moulding can consistently reproduce", what would you say consistently achieving 2-10um (microns) gap is? Magic? Fairy tale? Because LEGO as I said earlier is said to have 2um tolerances [1] over their decades of producing the bricks. Even a more conventional 10-20um (order of magnitude higher) still works.
> The article talks about manufacturing tolerances.
As shown by your quotes, the article clearly mentions tip clearance, and not manufacturing tolerances, which you are infering. The article doesn't characterize the thermal expansion the "super polymer" is expected to undergo under normal operating conditions[1]: something Lego doesn't contend with.
All this to say: Lego's manufacturing tolerances alone can't falsify Noctua's claims because they ultimately are different metrics.
1. I imagine the expansion rate of the fan blade radius doesn't correlate linearly with that of the shroud, so the tip clearance changes with temperature. With this constraint, not even Lego could make its manufacturing tolerances equal the fan clearance, which has to be larger if you want the fan to predictably work without jamming over a 40-degree temperature range.
I interpreted it as: with the nature of fans and the associated vibration/movement, some gap is necessary and this is the limit given the precision of injection molding.
Phrased differently: a 0.5mm gap is the minimum possible to also be able to account for the 0.1mm (or whatever) variation in injection molding.
> a 0.5mm gap is the minimum possible to also be able to account for the 0.1mm (or whatever) variation in injection molding.
The Noctua engineers definitely designed the clearances to perfection and accounted for the variation in the manufacturing process, I don't doubt that.
The article says "being off by a tenth or two suddenly becomes a problem", the 0.1mm you also thought of. But that's the point of contention, 0.1mm is the tolerance from bog standard, cheap injection moulding. The limit of consistent precision is in the single digit microns. Noctua doesn't need anything near that.
Unless working with that polymer is difficult and comes with higher tolerances, this is probably just a case of the article's author trying to pump up stats. To bring it more to the techie world, it's something along the lines of "130nm transistors are at the absolute limit of what EUV lithography can consistently achieve".
Because it's spinning blades among manufacturing tolerances you also have to account for the blades expanding when rotating at high speed, and possibly working with 40-50 °C air from the components
I don't think that's it. You're referring to tolerances specified in the design. The article talks about the tolerances the manufacturing technique allows, and this process is an order of magnitude better than this article says. The material used and the design of the part influence how much it deforms in practice far more than the injection moulding process itself.
In their own description of Sterrox® LCP they say it has "extreme tensile strength, exceptionally low thermal expansion coefficient, high environmental inertia and excellent dimensional stability". With such an advanced polymer any deformation in operation has to be a rounding error compared to the manufacturing tolerances.
I'd love more white, personally. I also don't understand the obsession with black. For me, black objects are very difficult to observe in detail, and that irks me.
Well, I have a bunch of lower-end black fans, some of them quite old, from before transparent cases were a thing. They're actually pretty much gray if I don't wipe them off.
Noctua's signature... brown-orange? Whatever that color is, it has the same issue. The blades are basically gray if I don't wipe them.
Haven't seen anybody start a gray craze, though. Though I have a grayish motorbike that also shows dust and dirt like nobody's business (it's a bike I use strictly on paved roads).
Silver is the ideal color for hiding dirt. I had a silver car once. Unless you drove it down a dirt road during a rainstorm, you basically never had to wash it.
If it were about performance and not marketing, they'd try to optimize for resistance to dust adhesion and resource consumption: energy, cost, durability, etc.
Why do you think they don't optimize for things like performance when they often win performance competitions against other vendors for both sound mitigation and airflow?
Do you know they have a specific high efficiency line?
Have you ever had a noctua fan fail where you think another vendor fan would not have?
I have them. They get dusty at about the same rate as a pure black fan (which also shows gray/brown dust quite easily). I need to clean mine about every 6-9 months to keep them looking good enough to "show off". I generally run a Winix HEPA filter in each room of my apartment.
I don't think matte white is worse than matte black in terms of showing dust. They both do.
It seems that common means something very different to you and I. Less than 5% of something is common? To me that sounds like the definition of uncommon!
Because it's a fan and I don't want to see it, and if I must see it, I don't want it to have any color of it's own since chances are very low that whatever color it is just happens to be the perfect addition to all my other posessions next to it.
I don't understand why anyone would think this is an obsession with black.
TIL: Generally all plastics exposed to UV start to photodegrade. If you google why old computers turn particularly yellow most sources point to bromine-based flame retardant agents in the plastic, but some people make a convincing case[1] that ABS just naturally turns yellow in UV light.
Not much real research into that topic, interestingly.
I've had a few experiences with retrobrighting and having it come out really nicely, then after being stored away in a box for a couple of years it's somehow yellow again. It's probably different with different plastics but it doesn't seem so clear cut that it's always the UV light causing it.
Not sure why all the fire retardants are needed. Besides, steel probably retards fire more effectively than most fire retarded resins and is probably far more recyclable.
In the uncommon event that something in your computer catches fire, the flame retardant keeps the fire from igniting the otherwise flammable plastic and potentially burning your house down.
Totally. I used to favor black a long time ago when most computers were still gray and the idea of having everything in black was really cool, but since realizing that details and controls are harder to discern on black, I’m all in on white and silver. It’s also less prone to showing fingerprints.
I always thought the grey Noctua Redux fans were their nicest looking offerring, despite being their lower end. I don't understand how they settled on that.
Maybe it is just my limited production knowledge, but wouldn't it be possible to injection mold a bigger part and then mechanically shave off the last few fractions of a millimeter using any number of ways? Tooling costs too high. But in the simplest form you could essentially spin the fan against some adjustable abbrasive to shave off the final bits.
Granted, there may be other places in which the molding precision may matter, which would make this an impractical solution.
The fan blades deform & vibrate under rotation & airflow. Controlling that deformation is the point of the use of materials like Noctua's Sterrox LCP or other flglass-fiber reinforced LCP materials in other premium fans. So lower clearance isn't just a matter of manufacturing tolerance
Okay. Seems like low noise is another big customer draw. So what's the difference for those measures between this difficult to manufacture fan and one with clearances that are easier to manufacture? If either is particularly significant, it's quite a bit more interesting than the measurement of the clearance.
Even if it is the case, and not simple an omission to focus the narrative, does it matter? Case fans pull what 4 watts? 5 watts? Who cares if it pulls 200 milliwatts more than a competitor when it's cooling a GPU and CPU that consume more than a hundred times what it can consume
That's really high. Like usually they are 100-150mA (so sub 2W) Lots of controllers would be 1A max.
The tolerances are for noise mostly. I'd consider the noise (and longevity) the single most important part of fans (else most fans can spin close to 3k rpm and cool)
Transcoding multiple video streams, running a VM and running various other tasks, mine rarely passes 10, and is currently sitting at 7 (only 2 transcodes at this time).
The question is not about saving milliwatts-hours on your electricity bill, it is about where these milliwatts are going.
One is heat, heat is not great, it puts more stress on components, mechanical and electrical, reducing longevity.
Another, maybe more important is noise. The power that goes into making noise is power that is wasted, noise is inefficiency, and reducing noise is an efficiency problem.
Tighter tolerance isn't universally a good thing. It might make the fan more susceptible to damage due to mishandling or dust. They might be selling a fan that has a shorter useful life for no real benefit.
As a physicist, it's not at all clear to me that tighter tolerances would lead to higher efficiency or less noise. I assume it shakes out in the CFD simulations, but I would be curious to know the explanation.
I thought the primary gain in efficiency came from the large blades, with the blade shape the next most important factor. Gaps between the blade and housing feels like a single-digit percent effect.
For the same reason a winglet is used on an airplane, or a ducted fan is more efficient than a propeller: there is a large pressure difference on the end of a wing or propeller, and the high pressure side will jump around to the low pressure side and cause a tip vortex. This causes an induced drag, which moves the lift vector backwards (as drag, but not skin friction drag). Higher aspect (think wind turbine blades or glider wings) minimise this, as do winglets or ducting.
You are talking about velocities 100x faster than the air in your chassis is moving. You might be right about the effect, but it seems so tiny in this application.
The specific fan in question has a rated max power draw of 1.8 W. In actual deployments it's going to be a lot less since ~nobody is running a noctua fan at 100% speed unconditionally
"In actual deployments it's going to be a lot less since ~nobody is running a noctua fan at 100% speed unconditionally"
I run dual 36w Delta fans at 100% in my computer case. I use the outflow as positive pressure forced exhaust for my enclosed CO2 laser, which itself has an ultra-weak venting fan.
It isn't that loud. A simple no box does the trick.
Yeah, but those aren't noctua fans. Noctua's claim to fame is being lower noise, not moving the most air. I'm sure somebody is buying a premium low noise-focused fan and then pinning it to max, but that's definitely not going to be typical.
which is why you went with Delta and not Nocturna I would think? Deltas are fine in an otherwise noisy environment but they’re misery in say a bedroom at night.
Case fans pull what 4 watts? 5 watts? Who cares if it pulls 200 milliwatts more than a competitor when it's cooling a GPU and CPU that consume more than a hundred times what it can consume
Yes, exactly. The high precision is marketing, not something needed in the product.
My understanding is that the precision is supposed to help with noise. Less turbulence, etc.
FWIW, in my setup (10th gen i5, RTX 5070 Ti in an old Define R3 case), the 12 cm Noctua G2 fans run quieter and have a much less obnoxious noise than the old P/F series, which wipe the floor with the Arctic fan I bought for a computer that lives in the basement and sounds like it's about to take off.
You lead me to believe that they are targeting a niche "audiophile" market and probably not a commercial market. The concern in the commercial market would be energy savings vs. capital expenditure. Some commercial spaces actually introduce white noise into spaces to increase occupant density.
They are targeting people who want nearly-silent fans for computing devices and will pay considerably higher than average prices for them. I have several of them, and they are vastly quieter than the competition. Wouldn't be worth it in a commercial space, but I want my house to be quiet.
In my experience fans from manufacturers like Arctic can be almost as quiet similar Noctua, but cost 50% less. The difference definitely isn't vast for most models, although admittedly there's more QC issues and variation than with Noctua.
A lot of Noctua sales come from their brand value. People put Noctua fans into their gaming PC's, use headphones while gaming on them, and then turn off the PC. You don't really need the most silent fan for that, but people buy them anyway for the looks & premium quality.
I do love Noctua's coolers though, I appreciate the well thought design, manuals and free upgrade kits when you upgrade your system to a new socket type. But for case fans I'll jut buy Arctic and save money, except for things like server systems that run 24/7 in my bedroom where noise and durability are top priority.
I want the quitest fans and whether they are 10 or 20 bucks is irrelevant. We are talking about tiny amounts of money here for something thats gonna run for 5y+
> except for things like server systems that run 24/7 in my bedroom where noise and durability are top priority
... which is why I only have a few of them, rather than replacing the fans in everything I own. But for the things that need them, there's just nothing else as good.
They target people that want quiet/silent cases, obviously not commerical, unless you're going after the long life/warranty service. Or you go for their industrial line.
Exactly this. Most of the time you get poorly researched articles (or nowadays, AI slop) about some topics only very remotely related to what the company actually does.
Here, the article is about something interesting that the company has expertise in (and even "insider info"), shows off that they do serious engineering, and is interesting to the target audience.
If I'm buying a 12V or 5V fan, it'll almost certainly be a Noctua. I don't know if they're the best, but they certainly seem to be among the better brands, and at something like $25 for a fan, they are certainly not overpriced enough to justify the effort of researching something better.
So whoever you are at Noctua, congratulations! This + the 3d model release are likely really paying off.
Noctua are pricey but they also provide service that is in my experience unmatched.
We have a few hundred of their coolers in use and I have never had an issue getting warranty replacements from them with fans. The process is simple and they ship out a new fan ( I have warrantied probably 10 - 15 of the fans)
Not all marketing is bad. Many of the beloved cartoons from decades ago were meant as marketing materials for toys and various kid items (i.e. lunch boxes). It doesn't mean it's automatically soulless.
In this case, I finally understand why they chose their most iconic colors, and appreciate the time they take on precision engineering.
With Noctua I highly doubt that is the case given their track record for quality overall and all other information available around their design and engineering process. As far as I know based on all the information I have seen all the design and engineering is done in Austria. They also have a track record of only releasing things once they are satisfied something performs within their standards. Something that would be next to impossible when solely relying on external fabs and process engineering.
They also utilize different manufactures afaik (historically Taiwan, but also China these days) meaning they need to have pretty solid in house knowledge and expertise to make sure different factories produces similar results. When they first started utilizing Chinese factories people noticed visual differences and were worried about that. But Noctua at the time claimed that they made sure that performance was still the same. A claim that was put to the test by various review outlets at the time (I want to say gamer nexus did a big piece about it?) and confirmed to be true.
Having said that, if you do utilize external factories you automatically are making use of their process engineering to some degree as well. But, and this is difficult for many people to understand, that isn't a binary thing either. You can entirely rely on the factory to basically do everything for you and just send feedback on iterations but you can also work closely with them and actually get involved in the process itself.
The other thing to consider is that while China is known for making cheap items for the American market (because that is what Americans want) they have become experts in the tool and die needed to make those cheap items.
If you want top class injection molding tooling / machines or process you are probably going to contract a Chinese company to do it.
Yep, if you're a big enough source of income for the factory you can basically do whatever you want, up to and including stationing your employees in their factory year round.
Ugh. This is peak AI influencer. It’s a 21 page AI-generated presentation and an AI prompt for:
npm i -g @musistudio/claude-code-router
So it looks like the person who set up this GitHub repository didn’t even make the software that does all the actual work. They also never mention the original author in the README. But, of course, their AI prompt includes instructions to open their homepage, so this is effectively using the AI prompt for indirect advertisement. The HN submission title is also misleading. I didn’t see any maths in the repo.
Having written more sed invocations by hand than I care to remember, please bin me in the mediocre camp.
Aside: The speed at which AI can spit out complex diagnostics is nuts. Par is usually half a second for a dozen complex shell commands tailored to the exact problem at hand.
I feel like those articles all look at the wrong aspect: once AI companies are forced to compete on price (i.e. in 6 months), then Google’s TPU is going to be a massive advantage that’s almost impossible for Amazon or Microsoft to replace.
Even bigger advantage Google has, in addition to TPUs and other stuff, is bundling. Bundle paid AI tiers with other products and entice users with clever pricing tiers, and you're golden. In fact this is what Google already does. Cheapest paid AI tier - 4$ and you also get two hundred gigs of cloud storage and photo backup and office and stuff. You want more storage - sure we have that, plus some higher AI usage limits on top of that, or maybe you want average storage but much more AI tokens? We have that too. Family packs? Sure. And so on. OpenAI doesn't have a photo editor or a music streaming or or funny reels network to bundle their expensive AI with.
Amazon has its own hardware in the pipeline, Microsoft would be stupid to not work on one, so best to assume they are at the very least a buyer of such a thing from AMD/Intel.
If the big cloud players slowly agree to compete on price then that means that OpenAI and Anthropic are suddenly in a much weaker negotiating position.
“When AI labs raise prices, big spending on AI could shift from a flex to a liability.”
because companies will need
“proof of productivity gains or metrics that show a clear return for all this AI investment.”
which in my opinion is simply not true. I haven’t seen any good study that showed AI to actually improve productivity overall. It massively helps in some areas, but then promptly gets stuck in others. So you still need an expert to guide it.
> I haven’t seen any good study that showed AI to actually improve productivity overall.
AI is overhyped, but on the other hand, I think it would be difficult to deny the significant productivity increases when used appropriately.
For some tasks, it's huge. Some tasks that I might've spent 8 hours on, I can do in 20 minutes. That's very real and huge.
At the same time, that's not the average that I experience. Some things are pretty much a wash. Others might be 2x or 3x faster which is quite nice, but short of the hype. And some things can be very clearly slower with AI. Also some things are more unreliable with AI.
We need to get to a maturity point where we realize it's just another tool. An incredibly powerful one for many tasks, yes. But it's not magically the right tool for everything and not always the right answer.
After I discovered how to use git worktrees in Codex to work in three conversations in parallel, I am able to build apps with a scope that simply was not realistic before.
You obviously are not reviewing the generated code in any detail before merging it. This is not sustainable for the project as it will grow to be too large for what it needs to be.
There was one feature/screen that Codex built in a single 5k LOC file.
It was still perfectly capable of developing the feature and it was working as expected.
I had it break it down into multiple files, but if I wouldn’t have seen it during the MR review, I would not have noticed. The large file did not seem to degrade the performance of the agent.
It would be interesting to discover how large of a project in KLOC an agent can continue to effectively maintain without messing things up due to the large size.
reply