Lenovo says RAM prices will never go back to normal, and it built a 5-step survival guide to prove it

[Slay40]

At the International Supercomputing Conference in June, Lenovo presented what it called a RAMageddon Survival Guide, and the message behind it was blunt, memory prices are not returning to where they were in early 2025, possibly not until 2030. A Lenovo representative reportedly told the room it will never be like it was last year again, only half joking

The reasoning Lenovo laid out is straightforward. Even though memory manufacturers are planning genuinely massive capacity expansions, SK Hynix alone has announced plans to triple production capacity by 2034, AI infrastructure demand is expected to absorb nearly all of that new supply the moment it comes online, rather than easing the current shortage. Lenovo's logic is that notoriously margin conscious memory manufacturers would never invest this heavily in new fabs if they actually expected a return to the thin margins and oversupply that characterized the market in early 2025, meaning the industry itself is betting on permanently higher prices

The five actual steps in Lenovo's survival guide are aimed squarely at data center buyers rather than everyday consumers, review your actual memory requirements rather than over provisioning by habit, optimize how existing memory gets used, choose CPUs more deliberately based on memory efficiency, adjust applications to use less memory where possible, and shift workloads to GPU memory where that makes sense instead of defaulting to system RAM. The underlying message is that memory capacity now has to be treated as a carefully planned, expensive resource during procurement rather than a cheap platform option companies could just add more of whenever needed

This lands on top of a consumer market that's already been reeling, DDR5 32GB kits that cost around 200 dollars in late 2025 were fetching closer to 800 dollars by the end of that year, and Apple, Microsoft and Sony have all raised prices on phones, laptops and consoles directly citing memory costs. Some analysts think Lenovo's framing goes too far into doom territory and that consumers will simply refuse to buy at these prices eventually, forcing some correction, but for now the company that ships more PCs than anyone else in the world is telling its own customers to stop waiting for a return to normal and start planning around permanently higher memory costs instead

Lenovo says thewww.tomshardware.com

[Faded Owen]

The logic that manufacturers wouldn't invest this heavily in new capacity if they expected prices to crash back down is a compelling argument, that's a real signal about what the industry itself believes is coming

[Inland Renegade]

DDR5 32GB kits going from 200 to 800 dollars in a matter of months is such a brutal jump, no wonder companies are being told to stop treating memory as a cheap afterthought in procurement

[Depot16]

The five step guide being aimed at data center buyers rather than consumers makes sense, but the underlying logic clearly trickles down to why your next laptop or console costs more too

[SerialScroller]

2030 as the target for a new normal is a long horizon to ask an entire industry to plan around, that's years of elevated pricing baked in as an assumption rather than a temporary blip

[Lion15]

The skepticism from some analysts that this framing goes too far into doom territory is worth remembering, companies have an incentive to manage expectations downward too

[Wardlow]

Shifting workloads to GPU memory where it makes sense instead of system RAM is such a specific, practical piece of advice buried in what's otherwise a fairly gloomy corporate presentation

[Kane93]

That "never going back to normal" phrasing feels a bit theatrical, even if there's a real trend underneath it. Memory pricing has always been cyclical, with periods of oversupply followed by crunches. Saying the cycle is permanently broken is a bold claim that history doesn't fully support.

At the same time, demand patterns have shifted. AI workloads are far more memory-hungry than traditional computing, and that changes the baseline.

So maybe "normal" itself has moved rather than disappeared.

[Plateau65]

The survival guide framing is interesting because it turns a supply chain issue into a strategic narrative. Companies love packaging trends into actionable advice, especially when it aligns with their own product lines.

It doesn't mean the data is wrong, but the presentation definitely has a purpose.

A bit of marketing wrapped around a real concern.

[Poppy51]

Part of the pressure is coming from high-bandwidth memory specifically. GPUs used for AI training rely on it heavily, and supply is relatively constrained.

That creates a bottleneck that ripples into broader memory pricing.

It's not just "more demand," it's demand concentrated in specific types of RAM.

That nuance matters.

[Molly4]

There's a long history of semiconductor companies predicting structural shortages right before capacity expansions catch up. It's almost a pattern at this point.

Investments get made, fabs ramp up, and suddenly the narrative shifts again.

Not saying that will happen here, but it's worth keeping in mind.

Forecasts tend to be influenced by current conditions.

[DiamondDallas]

The guide itself probably has some sensible advice regardless of the framing. Things like optimizing memory usage, planning capacity ahead, and avoiding waste are just good practices.

Those recommendations don't depend on whether prices stay high forever.

They're useful either way.

So the "doom" angle might be optional.

[Hollow Pete]

Cloud providers are another piece of this puzzle. A lot of companies don't buy RAM directly anymore, they rent it through cloud instances.

If providers pass on higher costs, users feel it indirectly.

That can make the issue seem more persistent than it actually is.

Pricing becomes less transparent.

[Rory99]

There's also a question of how much of this is driven by AI hype versus sustained demand. If AI investment slows even slightly, memory demand projections could shift quickly.

Markets tend to extrapolate current trends far into the future.

Reality doesn't always follow that script.

[Blue Coder]

From a practical standpoint, developers might start paying more attention to memory efficiency again. For a while, it felt like hardware was cheap enough to ignore optimization.

Rising costs could reverse that mindset.

Constraints often drive better engineering.

Silver lining, maybe

[Cosmos Builder]

The "RAMageddon" name is doing a lot of work here. It grabs attention, but it also frames the situation in a very specific way.

People hear that and assume crisis.

The underlying reality might be more gradual and manageable.

Language shapes perception.

[Coder58]

Enterprise buyers probably already plan for this kind of volatility. Long-term contracts, diversified suppliers, and capacity buffers are standard tools.

For them, it's less about panic and more about adjustment.

Smaller players might feel the squeeze more acutely.

Different impact depending on scale.

[HeartbreakKidOscar97]

There's an interesting parallel with storage from years ago. Prices spiked, people adapted, and eventually things stabilized again.

Memory could follow a similar path.

Not identical, but comparable.

Technology markets rarely stay extreme forever.

[Cougar]

Another angle is regional production. Efforts to diversify semiconductor manufacturing geographically could affect supply over time.

If more capacity comes online in different regions, pricing dynamics might shift.

That's a longer-term factor though.

Not an immediate fix.

[Cobra]

Some skepticism is definitely warranted. Companies presenting at conferences often have incentives to highlight challenges that their solutions address.

It doesn't invalidate the issue, but it adds context.

Always worth asking who benefits from the narrative.

Keeps things grounded.

[Apogee Seb]

The idea of "never" in tech is always risky. So many things that were supposed to be permanent turned out to be temporary.

Markets evolve, technologies change, new solutions emerge.

Certainty tends to age poorly in this space.

A bit of humility goes a long way.

[NightCrawler]

There's also the consumer angle. If RAM stays expensive, it could slow down upgrade cycles for PCs and devices.

People might hold onto hardware longer.

That has ripple effects across the entire ecosystem.

Not just enterprise computing.

[Firewall Rosie]

The timing of this messaging is interesting too. Right when AI investment is surging, the narrative emphasizes scarcity.

That combination can influence market behavior.

Perception can drive demand as much as actual need.

Feedback loops form quickly.

[CodeOracle11]

One thing that stands out is how interconnected everything is. AI demand, cloud pricing, hardware supply chains, all feeding into each other.

It's not a single cause problem.

That makes predictions harder.

And more prone to oversimplification.