21
votes
Buying a lotta RAM now, as an investment ... thoughts?
Just a passing thought, came up in conversation. I'm not talking about warehouses-full, nor even "retirement savings" quantities, but like, all the RAM you and your friends and family could possibly need for the next 3-4 years.
Pros, cons? Too late? Too volatile? Too ___?
In general terms, RAM is a commodity sold on the free market and is constantly priced by that market based on supply and demand.
In this case, it has already gone up in price to reflect both current conditions and expectations / knowledge about future conditions (reduced supply due to AI companies taking a large share of future production and the 'big 3' largely declining to increase production / increased demand due to panic buying by consumers).
The price will change further as more information becomes available and current expectations are either confirmed or refuted, but at this point in time, it is not possible to say whether that movement will be upwards or downwards. It may seem like there is only doom on the horizon (and the news lately is certainly doing its best to score clicks based on this angle), but no one really knows what events are in store even in the next 12 months which could move the price again.
Because of this, I would definitely not speculate on the price of RAM (or any other commodity). The only reason to buy RAM now is if you actually need it now.
If you're interested in buying RAM specifically for gaming and you already have 16GB of RAM (even reasonably fast DDR4), I'll just add that this should be enough to run almost all games today at pretty reasonable settings. I wouldn't be rushing to buy more RAM until there was at least a specific game I couldn't run satisfactorily with my current setup.
Scalping is generally not a good idea for investing. You're paying an insane markup, and you'll be selling it "used" so you'll have to price below market rate. Even if ram prices spike even further, it'll be difficult to recoup your investment and also pretty time/labor intensive.
Do you really want to subject yourself to answering hundreds of Facebook marketplace "Is this still available?" messages to make a few bucks?
If you think the prices of RAM will go even higher in the future, that means you think AI development will continue at the same pace and require even more resources. In which case, it's probably better and easier to just buy Nvidia stocks. (Edit: or RAM manufacturer stocks, as @stu2b50 mentioned)
I'm also not really sure how much "all the RAM you and your friends and family could possibly need for the next 3-4 years"? I don't think most people buy RAM sticks that often. Unless you also consider built-in phone and laptop RAM, the amount of RAM most people need in 3-4 years would probably be 0.
I would never directly buy a commodity to invest in as an individual. Too much hassle and liquidity issues. For major assets, there are financial derivatives. In this case, if you think memory will become big business, buy stock in the like three companies that make it (Micron, SK, Samsung). Micron is the more specialized out of those, so will have higher exposure to memory prices, for better or for worse.
Generally if you want to get into investing for a profit it's better to buy shares in the companies that make the thing rather than the thing itself. Don't buy RAM, buy stock in companies that produce quality RAM. If demand goes up big time, they make a killing and so do you. Don't buy gold, buy stock in companies that mine the gold. Don't buy US bonds, invest in US defense contractors, etc. Also be ready to drop any investments the moment they aren't returning value to you - don't get attached to things.
It's faster and easier to get back out of these investments again this way since you don't have to resell the product yourself. Always set a stop loss, and just keep moving it up slowly as the investment makes money. Eventually it'll stop making money and crash, which is when the stop triggers and auto-sells all of your shares. That's your profit. Never bother trying to get in at the bottom or top of anything - nobody can predict where the top and bottom are going to be.
Frankly I'd expect RAM to normalize pricing by summer. As chips go it is the easiest to produce, not like processor fabs for example. Supply will bounce back fast, chip makers know this game well, they've been here before.
The AI data centers are on borrowed time, this is the gnarliest tech bubble we've ever seen and it's overdue for a reckoning. Only reason it hasn't happened yet is because five tech companies are propping up US GDP by themselves right now and none of the investors want to accept the reality that AI is not working out as advertised.
They are still buying the line of bullshit coming from the tech bros about it. Those data centers are unnecessary, AI is going to live in a card in your PC not in the 'cloud' infrastructure. I'm sure they'd all love to lock AI up in data centers so they can meter access, bu that's not going to work for too much longer.
Generally, I agree with your operating principle here, but bonds are a completely different financial instrument, not a product. Investing in defense primes instead of government bonds doesn't really logically make sense.
So in addition to every other comment here I feel there’s an important first question not being asked:
Can you afford to lose every dime you spend on this if it doesn’t work?
I don’t think it’s a great play to begin with for reasons mentioned and not, but when doing a speculative investment like this the first thing you need is capital you’re willing to risk.
The basic first steps would be:
How much can I afford to lose if suddenly DDR6 comes out or no one ever buys from me?
Assuming I have some money that fits that, what would a modest expected return be on that money?
Are you willing and able to risk say 10,000 to maybe net 1,000 because when it’s all said and done you only get a 10% mark up? That beats an investment account but it’s far from life changing and the risk/reward ratios suck.
And then finally when that’s all said and done, could you have made more money by spending that time and money on something else (different speculation, savings, skills, etc)
Edit: oh and what timeframe are you willing to be invested for? What if it takes 3 months/a year to move anything? Can you afford to not be liquid for that time
Too many eggs in one basket, I would say. Things happen to stuff in storage: theft, water, falling, crushing, mystery, already came malfunctioning. Even if prices goes higher, think of it as paying for storage and insurance that when your family needs it, it'll arrive safe and clean and functional. Let them worry about the cost of malfunctioning pieces and storage.
This is a very time limited situation caused by short-sightedness*, greed and the AI bubble. Three years from now, if the current incumbents haven't gotten their shit together I'd wager China will already have entered and stolen away this market, providing enough supply to normalize prices. There's a clear, persistent, global demand after all, which means there is money to be made on the production side of things.
* If RAM scarcity limits people's ability to afford electronic devices, including computers, phones, cars, TVs, etc. this would ultimately impact access to AI too... They can't beam the chatgpt directly into people's brains.
That doesn’t really make any sense. If anything, this is the opposite of short sighted.
The three companies capable of fabbing memory have two choices in the face of sharp demand from data centers: they can rapidly increase production capability or continue with current production rates and potentially miss out on revenue by not capturing all of the demand.
If AI is a bubble, then they’re making the correct, and wise, decision by not increasing production for what would be a bubble.
Indeed, but there are powerful forces invested in keeping this whole ball rolling. In the unlikely event that - for whatever reason - no ramping up of RAM production capacity is necessary in the next couple years, which I find unlikely (despite hoping for the bubble to burst sooner rather than later myself) - that will mean demand has fallen off a cliff, which will mean investing in RAM as a private individual is still a bad idea.
RAM sellers aren't AI companies, so I don't think there's any lack of planning there. Most chip makers are nervous about increasing production (which takes years) because last time prices crashed, and they had a supply glut.
It's short sighted to shaft the entire consumer electronics market in order to make short term profit from a bubble. There's a balance here, a tipping point where the (currently) inferior chinese products will become a worthwhile tradeoff to a lot of businesses, which may have long term negative repercussions on the incumbents.
That seems unlikely to be a factor. Ultimately RAM is a commodity. There is a spec, and it needs to perform per the spec and that’s about it. That cuts on both ends.
For RAM manufacturers, there’s little reason to care about consumer sentiment. A tiny minority of consumers buy RAM directly and even they could not care less about who made it.
On the other hand, no matter how what they do now, if Chinese companies can produce RAM that performs to spec for less, then the current manufacturers are going to feel that market force regardless. The idea that consumers in this counterfactual would be like “oh, SK was so nice to me in the AI RAM shortage incident so I’ll spend more for worse RAM to thank them” is laughable. For this kind of commodity you set a minimum floor for acceptable performance and then sort by ascending price.
This doesn't read like you're discussing this with me, it reads like you're responding to arguments you read elsewhere. Sentiment?
Right. Say you're a TV manufacturer. You wish to manufacture and sell TVs. If you don't manufacture TVs, you can't bring in any revenue. Your TV requires RAM, and you have the following choices: The supplier who will sell you what you want for 10X, where X is what you paid last year, and the one who'll suggest something that performs worse for 0.5X .
Most consumers, as you have implied, are not experts. They aren't going to care if you (the manufacturer) lower your standards. They want to not have to wait a year for their TV as long as it works at all. Can the TV that has to be much more expensive in order to make up for the scarcity of a key component in it truly compete (in the consumer market) against the much, much cheaper one? They both show movies.
It's no different for manufacturers. You're going to get quotes from the three (or four, in a hypothetical where there are chinese entrees) manufacturers, and pick the lowest.
Let's say you make TVs. SK is supplying you, say, 4GB of DDR5 for $20. Your contract ends and you need to decide whether or not to renew.
In timeline A, SK throws you a bone by supplying you 4GB of DDR5 for $25 even though that's well below market price.
In timeline B, SK will price you along with everyone else for 4GB of DDR5 at $80.
The AI datacenter race peters out more or less, and market prices settle at $40 for 4GB. Now, a Chinese manufacturer enters and is offering 4GB of DDR5 spec-compliant memory at $30, looking to enter the market.
In both timeline A and B, as the TV manufacturer you'd pick the $30/4gb Chinese manufacturer. Why wouldn't you? So in timeline A, SK just lost money for no reason.
As a commodity manufacturer, you're going to price (with negotiations depending on size of order) on market price. Your product is definitionally perfectly replaceable.
This assumes the Chinese businesses are going to massively ramp up and oversupply RAM. All of the industry reports and rumors point to Chinese companies behaving the exact same way as Western companies. They also want to avoid a future supply glut.
New supply will take many years: https://www.tomshardware.com/pc-components/dram/chinese-memory-maker-cxmt-prepares-to-file-for-ipo-aiming-to-raise-usd4-2-billion-usd-to-take-advantage-of-tight-memory-market-company-lays-out-path-to-profitability-as-dram-demand-skyrockets-worldwide
Generally, when it comes to investment, the principle is "buy low, sell high". I don't think buying now, when prices are already high, is a wise financial strategy.
Hard advise against.
If you really wanted to do this, the correct way to make a profit from current RAM price volatility is to get in your time machine and buy up RAM 6 months ago.
Buying up RAM now, after prices have massively spiked, on spec that they will go up further (and enough further to make the amount of money spent and the risk anywhere near worth it) would, in my opinion, be a terrible gamble with many high-probability outcomes of you losing that money.
Reminds me of people who tried buying up toilet paper and got stuck with a storage unit full of it, unable to return it or move it. Note: I am not passing moral judgement on you for considering this with RAM. There is an ethical difference between those two situations in that toilet paper was an everyday need for every person during a crisis situation (though what a weird commodity to have a panic-run on), while people can go about their everyday lives without major impact if they can't buy RAM for 6 months or a year.
This feels exactly like the runs on toilet paper in 2020.
With the added feature where the toilet paper becomes worthless every 3-4 years when a new generation (DDR6 probably isn't too far away) comes out.
I've lived through plenty of computer component hiccups. Every so often there's flooding in Taiwan or a fire at a place that makes hard drives or a pandemic changes needs and suddenly fab capacity is overrun for awhile. They will never last, and betting on them is extremely foolish.
It's literally peoples' job to make calls on whether or not to spend hundreds of millions of dollars and years of effort to increase production capacity, which will destroy the value of any panic-bought stash, or ride out the short term blip and let it normalize again. And they have far more knowledge of how the market they operate in works, how their supply chain works, and general economic knowledge.
It's literally a lose/lose scenario where you're betting against an entire industry in being able to deliver their product, when they have a long track record of delivering a product on a massive scale.
The Framework Desktop with 128GB RAM is currently worth looking at, if you think you will need 128GB in the near future. You are buying 128GB RAM, soldered on in a way you can't upgrade, but still at last years prices. When their stock runs out, expect it to go from $2k to $3k. I would not buy it purely as an arbitrage opportunity however, as there doesn't appear to be a huge demand, and you run the risk of selling privately.
Perhaps look into index funds instead. You're probably too late to any stonk party regarding RAM prices.