Moore's Law dying in the next decade

[Pawan Fangaria]

On 50th anniversary of Moore's law, Gordon Moore was interviewed at his home in Hawaii's Big Island. He gives credit to engineers extending Moore's law for so long. And I concur with his thinking as I wrote in my article "7nm node is arriving..." that the whole world is inclined to let the Moore's law continue.

View attachment 13795

Dr. Moore himself says that the rate of progress that we have seen so far will not be there now and the Moore Law will see its end by the next decade or so. He remembers Stephen Hawking answering to a question on the limits to IC technology. Dr. Hawking said about two things as limiting - i) the finite velocity of light, ii) atomic nature of materials. Today, we are seeing these as limiting factors, exactly. Dr. Moore confirms that it's the technological barrier that will stop scaling, not the economical.

I concur with this, actually Moore's law is a hypothesis based on economical projection. So, while technological scaling will stop (i.e. we will not see more lower nodes), the economical packing of density such as 3D ICs, and bringing up other materials than silicon, photonics etc. will continue. However, if cost doesn't scale, then that can bring the economical end as well. That will be a big wall, where semiconductor industry can see a severe downfall, new products will not be there to replace older ones, and the same products may continue for longer than 2-3 years.

 

[lefty]

Well, Moore's law has being slowing down the whole time. Origonally, he said transister count would double every year, then he revised it to two years. Now, it's what? 3-4 years?

 

[Arthur Hanson]

Pawan, I agree with most of your statement, but not the part of the semi industry seeing a severe downfall. When we hit the size and economical barriers we will see a vast expansion of the breadth of the industry in applications. Listed bellow are a number of areas of breadth expansion, some that could be as large as the current semi sector is by themselves.

Medical, body control and regulation using SOCs and SOCs with mems, body monitoring systems, internal and external drug regulation system, replacement of ocular and all other nerve inputs including all five senses.

Automation and robotics (some with all five senses greatly enhanced), the applications here are limitless

Far further penetration of all transportation up to and including full autonomous operation

Semi filters built like large display panels. Desalinization, chemical separation

Sharply increased penetration of solar in the energy sector with large increases in efficiency

Increased use in heating and cooling everything. Also transfer of energy

Artificial intelligence

New types of sensors not even made yet

virtual and augmented reality glasses

The semi industry will not only not see a downfall, but large and maybe even accelerating increases for years to come, although like anything else it might not be on a linear scale. Uses will be only limited by the imagination and I'm sure others could add to my short and limited list.

 

[Pawan Fangaria]

Well, Moore's law has being slowing down the whole time. Origonally, he said transister count would double every year, then he revised it to two years. Now, it's what? 3-4 years?

Now, it's not even 3-4 years, may be 3 years of not more than 3 rounds. Moore is saying it will hit the wall and stop there.

 

[Pawan Fangaria]

Pawan, I agree with most of your statement, but not the part of the semi industry seeing a severe downfall. When we hit the size and economical barriers we will see a vast expansion of the breadth of the industry in applications. Listed bellow are a number of areas of breadth expansion, some that could be as large as the current semi sector is by themselves.

Medical, body control and regulation using SOCs and SOCs with mems, body monitoring systems, internal and external drug regulation system, replacement of ocular and all other nerve inputs including all five senses.

Automations and robotics (some with all five senses greatly enhanced), the applications here are limitless

Far further penetration of all transportation up to and including full autonomous operation

Semi filters built like large display panels. Desalinization, chemical separation

Sharply increased penetration of solar in the energy sector with large increases in efficiency

Increased use in heating and cooling everything. Also transfer of energy

Artificial intelligence

New types of sensors not even made yet

virtual and augmented reality glasses

The semi industry will not only not see a downfall, but large and maybe even accelerating increases for years to come, although like anything else it might not be on a linear scale. Uses will be only limited by the imagination and I'm sure others could add to my short and limited list.

Thanks Arthur! I agree on the breadth of use of semiconductor increasing in many areas as you listed. The advent of IoT will only increase that breadth of proliferation of semiconductors. And the depth of today's and may be another 2-3 rounds of ultra-nano nodes (7nm, 5nm) will also help in widening that breadth with lower area, performance, power.

But what I meant is that after 5nm in a deacde, the vertical wall will stop. As you have corrected me, breadth will continue to expand and we will not see the down fall there. But we are going to see down fall in another decade after that breadth saturates. Once, the breadth saturates, the semiconductor cycle (i.e. replacement of old products with new) will increase to may be 5 years?

 

[Arthur Hanson]

Pawan, I believe the largest market of all will be the AOE, automation of everything. It's already underway and has been for a while. Basically it's just the march of technology on steroids. From automating to simplest physical task to the most far reaching AI automating the thought process and combining physical and computational automation. This is already very large and just in its infancy. It will never saturate for it will be a key part in taking us off this planet and extending our reach into the universe. It is already playing a key part in lowering the cost of space travel. It will also allow for building mega buildings that are the sizes of cities that have the capability of being totally self sufficient. Our oceans are also largely untapped for space, food production and mining. Floating cities, static and mobile, offer a whole new realm and frontier for man. Already one can live on one of the mega cruise ships if you want, if you have the resources. I cut my teeth on reading at least a 600 science fiction books, maybe even over a thousand, and much is going to become reality, as much already has. The market is literally limitless, the only limits there are, are the ones we place upon ourselves.

 

[Pawan Fangaria]

Arthur, that's really a fictional paragraph If that has to happen, then definitely semiconductor is a must. If space travel becomes cheaper for everyone to travel, then definitely the cities can become floating. I guess, land will become cheaper, no one should care for. But I don't know how far we should rely on those lofty ideas. AOE or IoT is definitely realizable.

On a side note on space, NASA is doing a great job facilitating world wide competition from school level (class 10th, 11th) on Space settlement. There are great ideas coming up from students. This year my wife (a physics teacher) led team of 12 students from her school at ASIA level, they won that level. Now they are planning to visit NASA for world level competition. When I see their project (no time to look at that in detail though), it's really about how cities, as you say, can be established in space and they really work at estimating actual requirements, postulates, numbers.... To me it looks like fiction, but who knows it can be realized in next 50 years?

 

[Arthur Hanson]

Pawan, Space travel is the end game, but autonomous cars, busses, trucks, trains, ships, planes are to come and we are already seeing automation in these areas as with single pilot executive jets instead of two. Restaurants are just starting to use phone and tablet ordering and automating drink and bar functions. 3D printing of all types will require large numbers of semis. In the home, light switches, locks and climate controls, active glass coatings for privacy and temperature control all are areas that have low penetration now and are quite large. Gesture control of everything is another large area for expansion. I could go on, but the market for semis and semi technology is still in its infancy in many areas with large markets and we will discover more as the technology advances.

 

[Pawan Fangaria]

Thanks Arthur! Your points about automation are valid and there semiconductor will definitely play a crucial role. However, do you see an issue about economy? If all vehicles go driverless, restaurants without staff and other employments filled by robots, how much unemployment it will generate? Where will these people be absorbed? I believe economy also has its limits, today most of the regions we see overburdened with debts - Eurozone, Greece, China.... And in economic slow down semiconductor industry is usually the first to get affected.

 

[Arthur Hanson]

Pawan, I live and breath economics. Farming and food use to be the bulk of human endeavor, then manufacturing, then basic services. In energy we went from wood, hydro (from Roman to modern times), hydrocarbons, nuclear, geo thermal, wind and solar and are more diverse than ever.

Agriculture as we now practice it is obsolete. For counties like India I hope they make the jump to permaculture. Permaculture requires no fertilizer, herbicides, insecticides or fungicides and no expensive equipment and is labor efficient with production unmatched in quantity and diversity. Using algae we can produce up to 700 time what an acre of farmland can. Crickets flour is a very efficient ant productive protein. The tragedy of the misplaced marijuana laws is the hemp meal is almost the perfect protein besides providing oil and many materials and all on marginal land. There is a bright future if we grab it. Truth and transparency are key. Look at Singapore as a model for economic structure, it is but one way of creating a productive and fulfilling culture. There is room for many different cultures and economies, but truth and transparency is key. Dealing with failure is also key, generally the most successful people have the highest percentage of failure, but when the win, they are the big wins. In short they reach far and know when to fold and move on to something else and let go, this is a key skill set in itself. I could go on but space and time don't permit. Opportunity is limitless and luck is always a factor. If you want more specifics in a single area, let me know. Diplomacy from a personal level to a world wide level will be the most valuable skill set of all.

 

[Pawan Fangaria]

Arthur, These are great intellectual discussions. I know, each topic you described is a book in itself. As for semiconductors, that also has entered the lives of people like food and water in 30+ years of semiconductors, I wrote in one of my earlier blogs. In that I acknowledged that the semiconductor business may be slow or stagnate, but will not dip. However, as you say, it will develop stronger. For that, stronger economy is needed. Does that mean the debt burden will keep on increasing, or in other way money will move from other sectors into semiconductors? It might be possible that land becomes exponentially productive to reap the agriculture sector in itself!

 

[Staf_Verhaegen]

Dr. Moore himself says that the rate of progress that we have seen so far will not be there now and the Moore Law will see its end by the next decade or so. He remembers Stephen Hawking answering to a question on the limits to IC technology. Dr. Hawking said about two things as limiting - i) the finite velocity of light, ii) atomic nature of materials. Today, we are seeing these as limiting factors, exactly. Dr. Moore confirms that it's the technological barrier that will stop scaling, not the economical.

One other possibility for circumventing the laws of physics is do your computation in another way, like with qubits and quantum computers. The exponential technology improvement in traveling has stopped with the Concorde. It is now waiting for teleportation to make the next big leap in travel speed.

 

[Arthur Hanson]

Staf, I have been studying single layer planar technologies for over five years and started with grapheme. As I have posted on these pages in the past, some of these materials hold the promise of a 100 fold increase in speed at sharply reduced power. At a technical conference a year ago I asked Dr. Nalamasu of AMAT when he thought this technology will reach the market and he said five to six years. I'm in close touch with one of the top technologists in the world on this subject for many years and this may be speeded up by the first commercial use of grapheme maybe coming on the scene shortly. He said this is the opening of the door that may speed things up. If you have any information on this tech, please keep me informed. Sometimes I get to far ahead.

 

[Staf_Verhaegen]

If you have any information on this tech, please keep me informed. Sometimes I get to far ahead.

I have to admit I don't follow the technological side that closely anymore. The potential of graphene and other one layer techs has been shown by the academics but as far as I know the problem of being able to produce the blank wafers with low enough defectivity so chips with billion of transistors on them can be mass produced is not solved. I hope it is not a repetition of the carbo-nanotubes story which was the hype before graphene came on the scene.

 

[IanD]

There have been loads of fabulous new semiconductor technologies coming out of laboratories for the last twenty years or more which have promised to revolutionize the industry. The vast majority have failed because they couldn't be scaled into cost-effective mass production of huge volumes of chips with a huge number of devices per chip, which is what drives the semiconductor market.

If graphene (or anything else) delivers on the promise of an order of magnitude speed increase or power decrease, it still won't replace silicon unless the cost is reasonably close to it -- not an order of magnitude (or more) higher like GaAs or InP or...

Otherwise it will be restricted to small niche markets willing to pay a big price premium.

 

[Pawan Fangaria]

I agree with Ian. Moore's law needs both technology scaling as well as cost reduction on economic scale. And economic scaling is prime in that, silicon has been providing for so long. If any other technology provides higher speed and lower power, but does not scale economically, then that does not serve the purpose as stated by Moore's law.