Corsair redesigns DDR5 packaging to combat rising RAM scams

[Alfonso Maruccia]

The Scam Bubble: Memory prices are soaring, and scammers are following the money. With DDR5 and other DRAM products commanding record premiums, fraudsters are increasingly targeting memory buyers with convincing fakes. Corsair's latest packaging overhaul is designed to raise the bar, at least a little.

Earlier this month, Corsair rolled out a new packaging design for select DDR5 products. The shift reflects a market under strain: chipmakers are posting record revenues, while downstream customers are grappling with shortages, inflated prices, and a surge in fraudulent sales. Improving security and "transparency" around what's actually in the box has become less of a branding exercise and more of a necessity.

The ongoing memory crunch has produced some ugly side effects. Reports of predatory scams are rising, with unscrupulous sellers shipping obsolete DDR2 modules disguised as modern DDR5. At the same time, large IT players are absorbing much of the world's DRAM and NAND flash output, tightening supply and pushing prices even higher.

Corsair has been using its new package design since January, across all the Vengeance DDR5 memory SKUs sold in 2-module configurations. The design replaces the old cardboard boxes with a sealed plastic clamshell box, which is also made from recycled material. Furthermore, the new packaging provides a convenient degree of protection against electrostatic discharge.

The key change, though, is transparency. The clear clamshell allows buyers to see the actual memory modules before opening the box, making it far easier to verify that the contents are genuine DDR5 sticks rather than swapped or counterfeit hardware. Corsair says the design also makes legitimate returns simpler and more trustworthy.

To further deter tampering, the packaging includes a mid-section label that tears when opened. That makes it far more difficult for scammers to remove authentic modules, replace them with fakes, and reseal the package for resale.

Corsair notes that the stakes are high enough that even Light Enhancement Kits (LEK) – nonfunctional dummy modules meant purely for aesthetics – are being passed off as real RAM. With DDR5 prices soaring, the company argues that stronger physical safeguards are no longer optional.

The new see-through plastic box won't be used on every Corsair memory product. For SKUs that still ship in traditional cardboard boxes, the company says additional security labels will be included to provide a similar level of protection. In a market where a convincing box can be half the scam, Corsair is betting that seeing the hardware itself will make all the difference.

Permalink to story:

Corsair redesigns DDR5 packaging to combat rising RAM scams

 

[yRaz]

This ram shortage is a real pain, I wanted to upgrade all my AM4 systems to 128GB. I do think there is an opportunity for someone else to come in and start making RAM. Intel is having trouble filling all the spots on their 18A node, would be interesting to see them fill the gaps in their production to bring in some extra revenue

 

[Mark Fuller]

The so called RAM shortage is I think is actually a ruse to soak more money out of our back pockets. This is not the first time suppliers or manufacturers have created an artificial shortage just to get more for less. They save money by not spending to supply the product and then get more money out of every unit sold. Just who is to blame? Is it the merchant middle man or the manufacturer themselves. Or maybe a conspiracy among so many government bureaucrats and some CEOs in the supply and manufacturing chain and maybe products and services being sold in this country originating offshore overseas somewhere. God knows the poor buggers over there that work for slaves wages could use some more money. Terrible thing though, I never heard of them actually getting any of it.

 

[yRaz]

The so called RAM shortage is I think is actually a ruse to soak more money out of our back pockets. This is not the first time suppliers or manufacturers have created an artificial shortage just to get more for less. They save money by not spending to supply the product and then get more money out of every unit sold. Just who is to blame? Is it the merchant middle man or the manufacturer themselves. Or maybe a conspiracy among so many government bureaucrats and some CEOs in the supply and manufacturing chain and maybe products and services being sold in this country originating offshore overseas somewhere. God knows the poor buggers over there that work for slaves wages could use some more money. Terrible thing though, I never heard of them actually getting any of it.

Unfortunately, I think you're wrong. The US has turned the money printers on full blast and they are aimed directly at money furnace. It's forgien companies that make all the memory and they aren't buying our US AI hype. It would take years to build new memory factories and billions of dollars to do so and they don't believe the AI boom will last. They're more than happy to make money off of selling AI hardware for the time being, but they have so little confidence in the long term profitability of AI that they won't expand capacity only to be left hold a bag that brings the cost of RAM down once the bubble bursts.

This AI thing is strictly big in the US, hardly anyone else in the world cares about. The US talks about how it wants to be a world leader in AI, makes policies to prevent other countries access to AI hardware and those countries are saying back to us "we literally didnt want your AI hardware"

But as the long as the money printers are printing and the money furnace are burning, we won't see an end to the RAM problem. It's also kinda funny that almost half of all AI hardware is just sitting in warehouses or "dark" data centers that can't be turned on.

 

[JMGuy]

Unfortunately, I think you're wrong. The US has turned the money printers on full blast and they are aimed directly at money furnace. It's forgien companies that make all the memory and they aren't buying our US AI hype. It would take years to build new memory factories and billions of dollars to do so and they don't believe the AI boom will last. They're more than happy to make money off of selling AI hardware for the time being, but they have so little confidence in the long term profitability of AI that they won't expand capacity only to be left hold a bag that brings the cost of RAM down once the bubble bursts.

This AI thing is strictly big in the US, hardly anyone else in the world cares about. The US talks about how it wants to be a world leader in AI, makes policies to prevent other countries access to AI hardware and those countries are saying back to us "we literally didnt want your AI hardware"

But as the long as the money printers are printing and the money furnace are burning, we won't see an end to the RAM problem. It's also kinda funny that almost half of all AI hardware is just sitting in warehouses or "dark" data centers that can't be turned on.

I used to agree with this take on AI almost word for word.

I’ve been a huge skeptic of AI hype and argued against it pretty fiercely—even recently. And I still think most consumer-facing AI is borderline snake oil: flashy demos, shallow utility, and a lot of marketing dressed up as inevitability. For me, that’s still largely true when it comes to the slop companies like Microsoft and others keep bolting onto consumer products.

But something specific slightly shifted my perspective recently. I was awarded a contract to use AI to build extremely complex software prototypes for a U.S. defense contractor. No consumer focused slop. Hard engineering problems with real constraints—and effectively unlimited compute.

And I’ll be blunt: it increased my productivity by something on the order of 100x. And not in some vague “it feels faster” way, but in measurable output per day. Design cycles that would normally require multiple engineers working for months were compressed into days by a couple. Rapid iteration. Simulation support. Code scaffolding. Architecture exploration. When tuned properly, it stopped being hype and started becoming leveragable.

The catch, as I quickly discovered, is that tuning is more art than science. I think my specific background expertise allowed me to use it successfully in ways others were not able to do. Like any tool, user-skills matter.

Most people are using these systems incorrectly and applying them to the wrong categories of problems. They treat them like magic chatbots instead of probabilistic tools that require tight steering, validation layers, and deep domain knowledge. Used casually, they’re mediocre. Used deliberately, with some expertise in the how guiding them, the capability jump is real.

So I partly agree with you and partly don’t.

I absolutely agree that consumer AI adoption is wildly overhyped. I agree that outside the U.S., the cultural obsession with AI is far less intense. I agree that monetary distortion is clearly present. And I agree that some of the infrastructure build-out looks speculative and ahead of demand.

Where I hesitate is the idea that foreign manufacturers “don’t believe” in AI. The global memory and accelerator ecosystem—companies like Samsung Electronics, SK Hynix, Micron, NVIDIA, and AMD—operates on long capital cycles and strategic forecasting. Whether they’re bullish, cautious, or somewhere in between, I’m hesitant to think their decisions are being made on meme-timelines. These are multi-year, multi-billion-dollar capacity bets in an industry that cannot pivot overnight.

What I suspect we’re seeing is not disbelief, but capital cycle distortion layered on top of a technology that genuinely has asymmetric value in certain domains.

AI infrastructure right now is heavily influenced by government, defense priorities, and large enterprise spending. It’s been amplified by policy and, historically, cheap liquidity. That means capital allocation isn’t purely organic. But distorted capital allocation does not automatically imply nonexistent utility.

The “dark data center” argument is interesting. But overbuild often looks absurd before demand materializes. Fiber looked absurd in the early 2000s. Cloud capacity looked excessive before enterprise migration caught up. And In hindsight, those overbuild phases might have been messy, but they were not entirely irrational.

Where I strongly agree is that AI as a near-term consumer revolution is absurdly overstated. I don’t see a sweeping transformation of everyday life happening in any immediate future. The average person does not need a probabilistic co-pilot for most daily tasks. In fact, my experience with smart home tech proves that this is often quite undesirable once the novelty wears off.

But AI as a corporate, industrial, and military force multiplier? Based on my direct experience, that now feels quite different. In a serious engineering environment—where incentives are aligned, budgets are real, and experts are steering the system—it can meaningfully compress iteration cycles and expands what a small team can accomplish. That has some dramatic strategic implications regardless of whether or not it ever becomes a beloved consumer product.

As for the RAM shortage, I’m inclined to see it less as a narrative about belief or disbelief and more as a function of supply concentration, high-bandwidth memory reallocation, capital intensity, and the long lead times inherent in semiconductor fabrication. Memory fabs are not toggle switches. Capacity changes happen over years, not quarters.

I am still skeptical. Application matters. And I still think most of the hype-cycle is just financial theater.

But I’m no longer completely dismissive. How can I be? After seeing what properly tuned AI can do in a high-stakes technical context, I’d be very cautious about writing the entire phenomenon off as just another money-printer bubble.

Some of it absolutely is.
Some of it very clearly isn’t.

With that in mind, if other countries don’t want to compete in that space, that’s totally fine by me.

 

[yRaz]

I used to agree with this take on AI almost word for word.

I’ve been a huge skeptic of AI hype and argued against it pretty fiercely—even recently. And I still think most consumer-facing AI is borderline snake oil: flashy demos, shallow utility, and a lot of marketing dressed up as inevitability. For me, that’s still largely true when it comes to the slop companies like Microsoft and others keep bolting onto consumer products.

But something specific slightly shifted my perspective recently. I was awarded a contract to use AI to build extremely complex software prototypes for a U.S. defense contractor. No consumer focused slop. Hard engineering problems with real constraints—and effectively unlimited compute.

And I’ll be blunt: it increased my productivity by something on the order of 100x. And not in some vague “it feels faster” way, but in measurable output per day. Design cycles that would normally require multiple engineers working for months were compressed into days by a couple. Rapid iteration. Simulation support. Code scaffolding. Architecture exploration. When tuned properly, it stopped being hype and started becoming leveragable.

The catch, as I quickly discovered, is that tuning is more art than science. I think my specific background expertise allowed me to use it successfully in ways others were not able to do. Like any tool, user-skills matter.

Most people are using these systems incorrectly and applying them to the wrong categories of problems. They treat them like magic chatbots instead of probabilistic tools that require tight steering, validation layers, and deep domain knowledge. Used casually, they’re mediocre. Used deliberately, with some expertise in the how guiding them, the capability jump is real.

So I partly agree with you and partly don’t.

I absolutely agree that consumer AI adoption is wildly overhyped. I agree that outside the U.S., the cultural obsession with AI is far less intense. I agree that monetary distortion is clearly present. And I agree that some of the infrastructure build-out looks speculative and ahead of demand.

Where I hesitate is the idea that foreign manufacturers “don’t believe” in AI. The global memory and accelerator ecosystem—companies like Samsung Electronics, SK Hynix, Micron, NVIDIA, and AMD—operates on long capital cycles and strategic forecasting. Whether they’re bullish, cautious, or somewhere in between, I’m hesitant to think their decisions are being made on meme-timelines. These are multi-year, multi-billion-dollar capacity bets in an industry that cannot pivot overnight.

What I suspect we’re seeing is not disbelief, but capital cycle distortion layered on top of a technology that genuinely has asymmetric value in certain domains.

AI infrastructure right now is heavily influenced by government, defense priorities, and large enterprise spending. It’s been amplified by policy and, historically, cheap liquidity. That means capital allocation isn’t purely organic. But distorted capital allocation does not automatically imply nonexistent utility.

The “dark data center” argument is interesting. But overbuild often looks absurd before demand materializes. Fiber looked absurd in the early 2000s. Cloud capacity looked excessive before enterprise migration caught up. And In hindsight, those overbuild phases might have been messy, but they were not entirely irrational.

Where I strongly agree is that AI as a near-term consumer revolution is absurdly overstated. I don’t see a sweeping transformation of everyday life happening in any immediate future. The average person does not need a probabilistic co-pilot for most daily tasks. In fact, my experience with smart home tech proves that this is often quite undesirable once the novelty wears off.

But AI as a corporate, industrial, and military force multiplier? Based on my direct experience, that now feels quite different. In a serious engineering environment—where incentives are aligned, budgets are real, and experts are steering the system—it can meaningfully compress iteration cycles and expands what a small team can accomplish. That has some dramatic strategic implications regardless of whether or not it ever becomes a beloved consumer product.

As for the RAM shortage, I’m inclined to see it less as a narrative about belief or disbelief and more as a function of supply concentration, high-bandwidth memory reallocation, capital intensity, and the long lead times inherent in semiconductor fabrication. Memory fabs are not toggle switches. Capacity changes happen over years, not quarters.

I am still skeptical. Application matters. And I still think most of the hype-cycle is just financial theater.

But I’m no longer completely dismissive. How can I be? After seeing what properly tuned AI can do in a high-stakes technical context, I’d be very cautious about writing the entire phenomenon off as just another money-printer bubble.

Some of it absolutely is.
Some of it very clearly isn’t.

With that in mind, if other countries don’t want to compete in that space, that’s totally fine by me.

that's a very interesting perspective, I appreciate you taking the time to write it all out. I did over simply somethings and it isn't as black and white as I said, but Micron did publicly say they are not building any more memory fabs to meet demand.

AI is certianly useful and it's only going to get better, but it isn't the revolutionary tech we've all been sold on. If it was, the terms "AI" and "AGI" wouldn't exist. "AI" is used just as a marketing term at this point, noone knows what it even means.

I do see a crash coming, but that doesn't mean AI is not useful. I use it all the time in Excel, I help it automate tasks for me and I just want to make it very clear, I'm not 'anti-AI'. What I do see is a business model that is unsustainable and I think lots of the big money hardware companies see the same thing. Companies have been outright saying since last fall that they are warehousing components and now we know that many data centers can't even be turned on. Yet, we are still building more.

look, I work for a commercial construction company and we currently are building lots of data centers. I'll hear later on that the permits aren't ready or that they can't turn it on. I don't have access to the actual numbers, but I hear this stuff through the grape vine. I have reason to believe that perhaps over 50% of data center hardware produced in the last 12 months isn't even being used then why would Micron spend Billions to expand capacity.

We are at a point where people are willing to outspend their competitors to reduce access to hardware. TSMC, nVidia and everyone else are more than happy to pump their numbers while this is going on, but there is going to come a day when xAI, OpenAI, Microsoft, Google, Apple all have to stop because they're bleeding so much money with zero return.

AI isn't a bad technology, but the current business model behind it risks a global financial crisis.

 

[JMGuy]

that's a very interesting perspective, I appreciate you taking the time to write it all out. I did over simply somethings and it isn't as black and white as I said, but Micron did publicly say they are not building any more memory fabs to meet demand.

AI is certianly useful and it's only going to get better, but it isn't the revolutionary tech we've all been sold on. If it was, the terms "AI" and "AGI" wouldn't exist. "AI" is used just as a marketing term at this point, noone knows what it even means.

I do see a crash coming, but that doesn't mean AI is not useful. I use it all the time in Excel, I help it automate tasks for me and I just want to make it very clear, I'm not 'anti-AI'. What I do see is a business model that is unsustainable and I think lots of the big money hardware companies see the same thing. Companies have been outright saying since last fall that they are warehousing components and now we know that many data centers can't even be turned on. Yet, we are still building more.

look, I work for a commercial construction company and we currently are building lots of data centers. I'll hear later on that the permits aren't ready or that they can't turn it on. I don't have access to the actual numbers, but I hear this stuff through the grape vine. I have reason to believe that perhaps over 50% of data center hardware produced in the last 12 months isn't even being used then why would Micron spend Billions to expand capacity.

We are at a point where people are willing to outspend their competitors to reduce access to hardware. TSMC, nVidia and everyone else are more than happy to pump their numbers while this is going on, but there is going to come a day when xAI, OpenAI, Microsoft, Google, Apple all have to stop because they're bleeding so much money with zero return.

AI isn't a bad technology, but the current business model behind it risks a global financial crisis.

Thanks! I appreciate the construction-side perspective. Ground-level knowledge of permitting delays and idle hardware are more meaningful datapoints than most market commentary these days so I appreciate it.

For the record, my intent wasn’t to suggest you were anti-AI. Just adding adding my own context to the discussion—mostly because I myself have been a pretty hard critic. Takes a use-case for me to see the value. And the difference between bottomless-money AI capability and my personal pocketbook level is … other level.

At this point, I’m inclined to think that this is a big infrastructure wave that just isn’t scaling cleanly with capital running ahead of demand. That likely means some big winners and big losers as things smooth out over time. Capitalism in the U.S. is structurally more willing to overshoot than most countries I think; especially big-tech.

And I’m with you—a hard correction is likely coming. I guess I don’t see this, by itself, as something that structurally threatens the global financial system yet though. I’m thinking more likely a tech cycle reset than a 2008-style event.

Cheers

 

[Mark Fuller]

Thanks! I appreciate the construction-side perspective. Ground-level knowledge of permitting delays and idle hardware are more meaningful datapoints than most market commentary these days so I appreciate it.

For the record, my intent wasn’t to suggest you were anti-AI. Just adding adding my own context to the discussion—mostly because I myself have been a pretty hard critic. Takes a use-case for me to see the value. And the difference between bottomless-money AI capability and my personal pocketbook level is … other level.

At this point, I’m inclined to think that this is a big infrastructure wave that just isn’t scaling cleanly with capital running ahead of demand. That likely means some big winners and big losers as things smooth out over time. Capitalism in the U.S. is structurally more willing to overshoot than most countries I think; especially big-tech.

And I’m with you—a hard correction is likely coming. I guess I don’t see this, by itself, as something that structurally threatens the global financial system yet though. I’m thinking more likely a tech cycle reset than a 2008-style event.

Cheers

I used to agree with this take on AI almost word for word.

I’ve been a huge skeptic of AI hype and argued against it pretty fiercely—even recently. And I still think most consumer-facing AI is borderline snake oil: flashy demos, shallow utility, and a lot of marketing dressed up as inevitability. For me, that’s still largely true when it comes to the slop companies like Microsoft and others keep bolting onto consumer products.

But something specific slightly shifted my perspective recently. I was awarded a contract to use AI to build extremely complex software prototypes for a U.S. defense contractor. No consumer focused slop. Hard engineering problems with real constraints—and effectively unlimited compute.

And I’ll be blunt: it increased my productivity by something on the order of 100x. And not in some vague “it feels faster” way, but in measurable output per day. Design cycles that would normally require multiple engineers working for months were compressed into days by a couple. Rapid iteration. Simulation support. Code scaffolding. Architecture exploration. When tuned properly, it stopped being hype and started becoming leveragable.

The catch, as I quickly discovered, is that tuning is more art than science. I think my specific background expertise allowed me to use it successfully in ways others were not able to do. Like any tool, user-skills matter.

Most people are using these systems incorrectly and applying them to the wrong categories of problems. They treat them like magic chatbots instead of probabilistic tools that require tight steering, validation layers, and deep domain knowledge. Used casually, they’re mediocre. Used deliberately, with some expertise in the how guiding them, the capability jump is real.

So I partly agree with you and partly don’t.

I absolutely agree that consumer AI adoption is wildly overhyped. I agree that outside the U.S., the cultural obsession with AI is far less intense. I agree that monetary distortion is clearly present. And I agree that some of the infrastructure build-out looks speculative and ahead of demand.

Where I hesitate is the idea that foreign manufacturers “don’t believe” in AI. The global memory and accelerator ecosystem—companies like Samsung Electronics, SK Hynix, Micron, NVIDIA, and AMD—operates on long capital cycles and strategic forecasting. Whether they’re bullish, cautious, or somewhere in between, I’m hesitant to think their decisions are being made on meme-timelines. These are multi-year, multi-billion-dollar capacity bets in an industry that cannot pivot overnight.

What I suspect we’re seeing is not disbelief, but capital cycle distortion layered on top of a technology that genuinely has asymmetric value in certain domains.

AI infrastructure right now is heavily influenced by government, defense priorities, and large enterprise spending. It’s been amplified by policy and, historically, cheap liquidity. That means capital allocation isn’t purely organic. But distorted capital allocation does not automatically imply nonexistent utility.

The “dark data center” argument is interesting. But overbuild often looks absurd before demand materializes. Fiber looked absurd in the early 2000s. Cloud capacity looked excessive before enterprise migration caught up. And In hindsight, those overbuild phases might have been messy, but they were not entirely irrational.

Where I strongly agree is that AI as a near-term consumer revolution is absurdly overstated. I don’t see a sweeping transformation of everyday life happening in any immediate future. The average person does not need a probabilistic co-pilot for most daily tasks. In fact, my experience with smart home tech proves that this is often quite undesirable once the novelty wears off.

But AI as a corporate, industrial, and military force multiplier? Based on my direct experience, that now feels quite different. In a serious engineering environment—where incentives are aligned, budgets are real, and experts are steering the system—it can meaningfully compress iteration cycles and expands what a small team can accomplish. That has some dramatic strategic implications regardless of whether or not it ever becomes a beloved consumer product.

As for the RAM shortage, I’m inclined to see it less as a narrative about belief or disbelief and more as a function of supply concentration, high-bandwidth memory reallocation, capital intensity, and the long lead times inherent in semiconductor fabrication. Memory fabs are not toggle switches. Capacity changes happen over years, not quarters.

I am still skeptical. Application matters. And I still think most of the hype-cycle is just financial theater.

But I’m no longer completely dismissive. How can I be? After seeing what properly tuned AI can do in a high-stakes technical context, I’d be very cautious about writing the entire phenomenon off as just another money-printer bubble.

Some of it absolutely is.
Some of it very clearly isn’t.

With that in mind, if other countries don’t want to compete in that space, that’s totally fine by me.

Actually, I think ya'll are agreeing with my conspiracy theory. Like yRaz said, it is the US government bureaucracy that is printing money and like I said, the problem originates offshore where you said they don't care about any AI problem that is mostlly a US concern. I am, however, glad to here that you were able to increase to increase productivity and save money. I am assuming for that reason you must be some kind of able and efficient techinian of sorts to do so. That would leave most of the others out in the cold, which could just be the very problem in the first place.