It isn't a question of mastering a 20-year old technology.
Yes it is. That's EXACTLY what it is.
at the cutting-edge where TSMC and Samsung are producing chips... based on 22 year old technology.
Some mildly amusing stuff done recently in tweaking a few things in the range down to 7nm... covering improvements in construction techniques for electron highways reduced down to 70 electrons wide.
But, then what ? China might see good news... as it suggests, by spending trillions in the effort, they might be able to catch up at some point in the future... perhaps even while others are still stuck in idle, waiting at that 7nm scale traffic light ? IF THEY ARE stuck there now... which isn't at all certain... given that the 22 year old technology seems it isn't giving up its secrets all that easily, even at the end of its life? The perennial premature proclamations aside, it isn't dead, quite yet ? So, how can you know what else is out there... what other innovations are lurking... perhaps just as opaque as the recipe for the secret sauce that makes that old stuff work ?
Hmmm.
FWIW... I bought a few shares of Qualcomm more than 22 years ago... under $1, so must have been '92... which sounds about right. Was well before 1998, and the rocket ride in '99 ? Pretty sure it was one of the key interests I posted about, here on SI, during the "soft open" before August '96 when SI first opened the doors to anyone and everyone who wanted to play ? Back then, before "the reset" had much SI history evaporate... the "invitation only" crowd on SI were all certified tech geeks and CEO's... mostly didn't know who it was you were talking to... and it doesn't matter now. QCOM's business was quite a bit different then, too... they were massively boring, but profitably selling "parts" for comm links... LAN connectors and stuff... while they waited on patent issuance, and license deals when the IP got validated. Today, they'd likely be dismissed out of hand as being "just another generic parts supplier, and an evil upstart of a patent troll" ? So, the point... quite a massive time warp is in play here ?
Since way back then... curiously... no one in the industry seems to have had a clue about, much less a plan, for building electron highways much below the 7 nm limit? Just been riding that same old horse ever since 1998... even knowing the glue factory is just around the corner ? Which doesn't mean there isn't something else waiting in the wings ? Just leaves a question: When you do master the current technology... how far behind are you ?
But is it true that the 22 year old technology solution... which it seems "shall remain nameless" even now... will run out of steam at that point ?
Yes... it does seem so. Maybe. The 22 year old technology that has been sustaining Moore's law on increasingly expensive life support since the last time it stroked and died, back in 1998... does reach some limits in physics at that point... barring new innovation.
So, that's it ? Yeah. Doom and Gloom... "End of the road" for Moore's law... AGAIN... exactly as it was back in 1998... ? Same thing is said in a steady drum beat every year... since before 1998... and still now ?
2013: End of Moore's Law: It's not just about physics
Feb, 2020: We’re not prepared for the end of Moore’s Law
Oct 2020: How did Intel lose its Silicon Valley crown?
Why does it even matter ?
Beyond historical interest, or the ephemeral issues in the competition between companies, or countries... it matters because this article is EXACTLY wro ng... in considering the economics of things. Chip prices... or the costs of making them... don't matter at all to consumers. Reality is... a decent desktop computer, today, costs about what a decent desktop computer cost in 1990... while a decent basic cell phone today... basically no longer exists... as the hot, new and expensive 2G phones from 1990 were not even smart phones... but, in the same time, the cost per cell phone minute went from over $1 per minute then, to... who the hell even knows what a minute of cell phone time costs anymore ? How much data does your plan come with, how many phones are on it ? No real comparison to make between phones in 1990 and today ?
What's changed isn't the cost or price of the devices, which has stayed remarkably steady for "new" stuff... but the "per unit of value" costs in "provision" of chips and services (WAY down), and the realized value in the utility of the functions (WAY up). In 1990... you could play computer games... while the (super hot chick) girls in the typing pool were all being re-trained to use word processors and the new accounting software... but, otherwise, not much really changed next year, with the upgrade to Windows 2.1 ? There were an odd few articles, but not many journalists who were obsessively writing articles annually flogging the dire consequences in the looming death of... the typing pool... ??? But, the typing pool now is well and truly dead. Moore's Law killed it. So today, your (randomly sexed, lookism and gender bias free, fully diversity compliant ) "assistant" has as much productive power in an hour... as the whole pool of (super hot chick) personal secretaries and typists had back then, in a week... Who Still Buys Wite-Out, and Why?
What's the economic impact of that ? Today you have the typing pool in your pocket, sharing space with your accountant, stockbroker, banker, and bookie... and probably the pizza delivery... who themselves are mostly digital creatures now, not people earning paychecks. Now, you can receive, read, sign, and return digital documents in a few minutes time, entirely obviating printing anything, or the need for expedited mail services, coming and going. Minutes and pennies, versus days and dollars ? Plus, there's all the cute kitten videos... and your company is on the verge of selling your corner office, banishing you to work, at much lower cost, from your lawn chair at the beach ? Does it matter that the new 5G might be a tiny fraction of a second faster, for some ? Probably NOT much of the same scale in impact, there...so, now at the cusp of a point in diminishing returns ? But, who knows what you'll soon "need" your next generation computer or phone to do with that greater capacity, that we haven't even begun to figure out yet ?.
Hint: the impact is not "hyper-inflationary"... but deflationary. Technology masks the driver, and also localizes and isolates the impacts in any inflationary tendencies that emerge. We've had a massive hyper-deflationary environment imposed on us by the technology, with massive per-person potential productivity increases, since the point computers began adding more value, in function and efficiency, than they cost us in time wasted re-booting, or re-loading Windows from scratch, after another Bill Gates produced BSOD event... that immolated your hard drive... right before you printed out that report to meet your deadline.
Tides, and the currents they create...
What happens next... with change occurring in that now 40 plus year long technology driven secular deflationary trend... that drives the dominant current we float along on unawares... in a sea of massive, innovation led productivity enhancing but deflationary drivers... What happens when that silicon fueled magic suddenly stops flowing a perpetual stream of more for less ?
I don't know. Maybe ask Siri ?
Sense: "Siri, when Moore's Law ends, what happens to productivity and inflation ?"
Siri replies: A single technology trend does not alter productivity and inflation in a vacuum. Whatever the density factor of transistors on a chip, improvements in AI and robotics will increasingly shift innovation to enable eliminating manual labor costs, reducing the value of labor, while also redirecting primary focus away from exploiting low skilled labor en masse in repetitive tasks. The very early stage $250 Roomba 600 series robot has, in the United State alone, already incrementally eliminated 650,000 job equivalents for house maids and janitors, each lost job opportunity considered worth between $22,619 and $40,904 per year, leading to unemployment spiking among the less skilled, while almost making iRobot profitable.
(Sense grumbles: No one ever fact checks Siri: iRobot Stock Spikes as Roomba Vacuum Cleaner Sales Soar Amid Shutdown Sense continues grumbling: How does being at home more, because of Covid, justify people who can't afford it... spending more money to NOT sweep the floor themselves ? When Spending Money Equals Happiness: Why I Bought A Roomba Continues grumbling. )
(Siri, not done yet, ignores Sense grumbling:)
Productivity effects are roundly ignored by monetary theorists... who are gleefully trying now to reinvent money functions in light of unexpectedly absent realization of inflation risks, which appear to have allowed "some greater degrees of freedom" in monetary theory than they can explain, while continuing to ignore drivers of and changes in productivity... with the productivity effect expanding and the monetary impacts amplified... as application of the basic technology grows and deepens in reach. Redesigning money with new functions "liberated" to ignore suppressed inflationary tendencies, while they are suppressed by secular deflationary trends... accelerates other pre-existing tendencies, particularly expanding "wealth transfer effects". The bankers (allowed to create digitized Roomba-dollars... that suck up any loose change that might be left in the economy following the deflationary impact when housemaids are put out of work), transfer the resulting change in value directly to themselves in the form of an expanded currency issuance, done as a counter-balance, without causing inflation in the result. The sustained benefits of productivity increases are thus made subject to the wealth transfer effect... so the bankers continue to pay the monetary theorists quite handsomely to ensure they remain confused on the subject of productivity, while they work to enable greater efficiencies in the design of schemes facilitating the transfers. However, when secular trends in the deflationary impact of productivity improvements suddenly falter, or when that trend comes to an abrupt end... the entire scheme, including monetary innovations, and wealth transfers, will fail and the economy will collapse, or it will suddenly revert into a hyper-inflation, until there is a functional "equal and opposite" adaptation made to the new secular trend.
Sense: That's enough Siri, stop.
Sense: Siri, is all of that inevitable ? Will robots take over the world ? Is Moore's Law ending ?
Siri: Maybe.
Sense: Siri, deploy the Roomba....
A soft whirring noise comes from the next room... The dog grumbles and goes outside. |
