Hacker News Reader

Comment by adventured | Hacker News Reader

Cool technology... however this:

"He said NRAM offers appealing cost benefits as well. Because of the small size of the carbon nanotubes, it can be easily shrunk, which means NRAM doesn't face the same manufacturing challenges as other forms of memory"

is a pretty ridiculous claim for the next decade or so at least. It should have said: theoretical cost benefits at some point in the distant future. Today NRAM has horrible cost negatives, because you're manufacturing with carbon nanotubes. It doesn't face the same manufacturing challenges as other forms of memory, it faces drastically greater challenges. It's going to take a long time to get this to an economy of scale, such that you can buy a gigabyte of NRAM for a competitive price.

View on HN

I work as a mechanical engineer designing CVD (chemical vapor deposition) equipment for research applications. Some of my systems are designed to grow carbon nanotubes, although admittedly not at the purity or wafer size that would be required for serious commercial semiconductor work.

My experience is that CNT systems are among the easiest jobs I get. They tend to have very simple gas systems, with only a few gases, and few other complications in terms of temperature or other complications.

This isn't to say that one could take a current chip fab and start running a CNT process tomorrow, there is of course going to be a lot of research and development still required. On the other hand, it certainly seems reasonable to expect that this won't be particularly harder than developing some other new process based on a different technology.

Looking at [1], the price of low-volume of single gram of high quality carbon nanotubes is $250. Volume prices go further down.

According to [2] carbon nanotube layer could weight 0.1 gram/square-meter.

So basically we'll spend $25 for a square meter of chips(more precisely dies) - that's fractions of a cent per gbyte.Pretty insignificant.

[1]http://cheaptubes.com/carbon-nanotubes-prices.htm#Single_Lay...

[2]http://nextbigfuture.com/2008/02/large-sheets-of-carbon-nano...

The price of raw silicon is also extremely low, and just like silicon, the processing is the expensive part.

Sure the processing is the expensive part - but his requires regular chip processing("cmos process") as far as i understand.

Nantero is not the only company making nanotubes. Nanotechnology as a whole, and carbon nanotubes (CNT) as a prime component, have been a hot topic in material science and chemistry for almost two decades now. Not just research for applications, but for production. They are not alone in trying to solve problems centered around nanotech. But the CNT production overall is still very low (only 5k tons in 2014 according to this article http://globenewswire.com/news-release/2015/04/22/727078/1013...)

No one is saying the iPhone 8 will have NRAM. The manufacturing research and implementation cycle is painfully slow. But if the cost/benefits may pay off in 2025, someone needs to be looking at it today.

What's the point of coming out of stealth at this time then?

I completely agree that it's a future, potentially viable technology, and it should be explored.

Some people in this thread certainly seem to think it's imminent, because they're claiming it's all ready to be mass manufactured due to compatibilities in the manufacturing process, and just taking Nantero's hype at face value (like we were all supposed to be drowning in memristors by now, just ask HP).

Not really, a bit further on in the article the founder mentions how they spent most of their time on the tooling to make it compatible with existing manufacturing methods. It should be cheaper than DRAM and slightly more than NAND.

That's impossible today.

Are you telling me they're going to be manufacturing millions of units of carbon nanotube RAM at $50 for 8gb in the next five years?

The product they're looking to sell is extremely small amounts of NRAM, at extremely high prices vs DRAM. What you're not going to see is $0.01 per mb of NRAM in the near future.

I can't tell whether you think millions of units is an insignificantly small amount of volume, and as a result the pricing for carbon nanotube RAM won't come down in price.

I would think they are targeting units in the hundreds of millions, if not billions of units over the next five years to get to scale.

The the lower power opportunity makes it an obvious choice for mobile devices (watches, tablets, phones) if the technology proves out.

Precisely, the article also states that they aren't aiming to manufacture, but to license the technology.

It sounds like they want to get this out in the marketplace. If we see two to three manufactures that can bring tooling up, and if they don't collude on price, I would think the consumer costs should come down due to the competition.

I believe the likelihood that we're going to see $50 / 8gb type prices from carbon nanotube RAM in the next five years, on the shelf at Best Buy, is zero. I don't believe they're (the companies that license it) going to be able to manufacture millions of high capacity units (multi-gb) at competitive cost, because they're not going to be able to produce it cheaply any time soon.

This is just another in a very long list of proclaimed traditional RAM killers that has been touted, only to come up empty on price or performance. In this case, it's going to be price and capacity. In my opinion this product will turn out to be far more difficult to manufacture at volume than what they're admitting. Five years from now, you still won't be able to buy it at scale for a competitive price.

For them to enable the manufacturing of billions of units of a carbon nanotube product, would practically make them wizards of chemistry. Nowhere on earth are carbon nanotubes being manufactured like that, and there's a good reason for why that is.

The first target market for this type of memory won't be best-buy retail units, but instead embedded in low-power/small battery devices. Smart Phones are where I'd expect these to first make an appearance - so 10s of millions of units over the next five years is a pretty inconsequential target, particularly if they are just licensing their IP.

They're not making or selling anything. They're only licensing the technology.

Indeed. I don't expect this to be any less expensive than SSDs were compared to regular HDDs when they first came out. But if they are lucky, NRAM or other type of next-generation non-volatile memory will face the same growth path as SSDs and a decade later will enter into mainstream devices.

Indeed. Getting the nanotubes to stick to the pads you want them to stick to is pretty hard from what I've heard and last I heard good defect rates were still double digit percentages. Maybe they've got some sort of technique for dealing with that or maybe they've solved the problem in some special case suitable for the regular arrays you'd find in memory but I'd be really curious about that.

Reply on news.ycombinator.com