Get unlimited access to Bloomberg.com for $1.99/month for the first 3 months: www.bloomberg.com/subscriptions?in_source=UKpostsOriginals

After spending 20 years as a salaried manager in corporate America I can truly say that management is the root of Americas lack of innovation. The majority of managers are focused on meetings, talking on the phone, and working on networking (talking to people for no reason).

The two brightest flames of US advanced manufacturing (Intel and Boeing) have recently made such massive and historic missteps in the last decade. It's really sad.

The real reason the US and all western nations missed out on this wave is because corporate heads wanted to produce their products in a cheap environment to maximize profits. Now all of the supply chains and the skilled workers are in Asia.

This is what happens when a stock price/profit focused CEO controls a tech company.

Morris Chang, the founder of TSMC was a director at Texas Instruments, he was passed up for promotions to less capable peers and saw the glass ceiling for ethnic Chinese. He went back to Taiwan to set up TSMC and the rest was history.

Intel: spends $4bn developing the tech but doesn't want to spend $0.2bn buying even a single development unit to use the tech

ASML had the forsight to buy the US company who pioneered EUV. Intel could have bought it. Applied Materials could have bought it. None thought it was worth the effort.

As a person who used to work in the semiconductor industry for decades: In a society that only believes in laissez-faire capitalism, it's only corporate profits that matters. That leads to all sorts of out-sourcing to increase corporate profits. Everything moves to cheaper locations that can do the work. Other issues like national (or world) security & stability don't mean much to corporations who are slaves to their quarterly returns.

What US did to Japanese semiconductor industry is really disgusting….RIP Toshiba Semiconductor

Regarding the chip shortage: For 4 years, I worked at a Texas-based semiconductor fabrication plant. (DUV and I-line) Our fab has been running 24/7 all throughout the pandemic. The bottleneck problem is, all 300mm wafers have to be put on a train, then a cargo ship or a plane, shipped to either China, Taiwan, Philippines, Japan, South Korea and Malaysia to be cut, then put into their final chip form, then sent back the the US, where they have to be redistributed, then installed wherever they are needed. Some are even installed on their circuit boards after this process, due to the circuitry ALSO being created overseas. There is no facility in the US, that can do this final process in a high capacity scale, nor plans to do this, because it’s cheaper to outsource to other countries. Not forgetting the fact that the machinery needed to do these complicated procedures, are all patented and built by Asian and Norwegian countries. Now, think about it pragmatically; If your country was just getting back to work from the pandemic, who do you think they would prioritize? Themselves or US (who had a president that downplayed Covid, blamed China for it in the first place AND by proxy, caused horrible crimes against citizens of any and all, Asian decent) ? I say that only because right after we had gotten shipments caught up after Trumps “tariff war” debacle (which is a tax that WE, the consumers pay) then we got hit by the pandemic. Once gain, we were having to store excess wafers due to the supply chain being backed up. We have $millions of dollars worth of chips that are STILL waiting to be sent overseas, finalized, then sent back. (Then you have factors such as excess shipping containers that are overcrowding Asian and US ports, making organization more difficult, a decrease in truck drivers, the occasional container that falls off the ship in a storm, excess fuel costs slowing down distribution, etc. Some higher end chips are flown over on cargo planes, which has also increased costs due to fuel and pilot shortages.) Y’all want faster chip production? Find a way to move final chip production stateside!!! I left that industry after witnessing poor management and poor future planning.

The US really woke up late. ASML bought a stake in the Zeiss subsidiary that makes these crazy mirror focusing systems. ASML only paid a billion dollars for that share. Even just a billion can do great things if you know how to spend it.

As a Taiwanese involved in semiconductor industry, 5:24 is a very misleading interpretation. Intel has a very different business model than TSMC. It’s a very consumer-facing company, doing both designing and manufacturing. Its profit ties heavily on its capability on launching new product to the market. TSMC only specializes in manufacturing and doesn’t launch any product. You can even say that TSMC’s chip design capability has fallen behind Intel for at least 50 years, in the sense that it doesn’t design its chip at all. Intel trades part of their chip manufacturing capability to chip design business. The differences in business model shouldn’t be interpreted as what the video intended it to be.

Bloomberg is interviewing itself these days?😂😅

Intel didn't use EUV because they tried to advance past 14nm but got stuck big time, but were too proud to switch tack and either buy from TSMC, or adopt EUV because unlike smartphone ARM chips where smaller nodes mean big power savings and longer battery life, x86 desktop/server chips doesn't need to be that power efficient at higher cost and lower profits. And also because they were facing minimal competition from AMD so could afford to get stuck for years, that is until AMD fought back with Zen/Epyc. If Zen/Epyc didn't exist Intel might still be on 14nm.

TSMC first overtook intel in 2016 at the 10nm node, which wass done without EUV. TSMC's first 7nm in 2018 (similar to Intel's old 10nm) was also done without EUV. EUV wasn't the biggest reason intel fell behind, but why it stuck on 14nm without back up plan for so long still puzzles me to this day.

America first time experiencing how it is when they make an advancement and don't capitalize on it, lol

We privatize profits and socialize costs in this country. It's not surprising that Government Sponsored Research results and support for technology gets sold out, the same way we offshore manufacturing. Another self inflicted injury.

"One atom thin layers"...wow. Human tech is inching towards the physically possible?

Intel CPUs has been offering IPC gains

USA did the same thing with Middle East oil in the 1970's. Shifted from a net exporter to net importer because foreign oil was "cheap". Then embargoes happened, wars happened and oil became a method for adversaries to attack the USA. Today we are the world's largest producer again but it took decades to fix that mistake.

Intel's ceo blew it.

I spent over 25 years at Intel and I have a slightly different perspective. Intel has a "copy exactly" philosophy when it comes to their fabs. All are identical. Intel simply couldn't get enough EUV machines to handle their leading edge capacity needs and they couldn't run different versions of their process in different fabs. So they chose to use a different lithography technique across all fabs. It's called multi patterning and it didn't work so well. Their mistake was in not running what's called a small "boutique" process line to get experience with EUV until they could get sufficient machines. Also the chips TSMC was making were much smaller is size and therefore they could take the hit of lower yields. The smaller the die the higher the yield at a given defect density. So they could get up to speed faster and still ship volume to their customers. It was a hard lesson for Intel but I think they are on their way back to process leadership. Only time will tell.

"We don't have to take over the car, we have the fastest racehorses in the stable, our competitors will never catch up with us. EUV is like the internet" - Brian Krzanich

Sickening that government grants are needed when they should have reinvested in the company. Just 🤒

Nothing like government funding when capitalism needs to win

Our early tech companies were started and run by engineers. Over time the engineers running these companies were replaced with bean counters. American workers were laid off and the work was moved overseas to be done at a much lower coast. It was just a matter of time. Blame it on the greed of the top level managers at the expense of the American worker.

That transitioned from 'America' to 'Intel' pretty swiftly

Having the machines here in the US means nothing. Almost no one here in the US knows how to run them. We need to invest a lot more in education.

nvidia doesn't even make chip, they just buy them from TSMC, why they market cap goes up was because of speculation on AI. it has nothing to do with chip manufacturing. in fact nvidia only grew because they primary competitor in the past, 3dfx(dead) and ati(now amd), spend more of their money building part of their hardware, while nvidia focus on solely design and software. demostrating the superiority of the outsourcing model. AMD would follow this selling their fab business(now GlobalFoundries) and focus on design and software as well, netting them the deal with sony playstation which is also seeking to get out of making their own hardware. the entire TSMC business is build on outsourcing chip production, if the outsource model isn't superior, it would not have been so dominate, yet it is also misleading to amusing outsource number are the entire chip industry, obiviously intel and many chinese companies that build inhouse are not accounted for. I have yet to see any media put up a decent research that compare both outsource and inhouse production. without which you cannot do a comparison between nvidia and intel as it would be worst than comparing orange to apple.

It started when we outsourced everything to get cheap labor when I was a kid in the 90's. My grandparents worked for HP. They sent everything overseas. That was the worst, most corrupt choice. Now we all pay.

Same thing happened with GSM mobile technology. Invented by USA but realized in Europe

America's problem is not worlds problem.

By 2027, TSMC, Samsung and Intel new plants will be operating. My question is - who will be buying those newly minted OVERCapacity chips?

Just like Boeing, Intel was run by market people, not engineers. Instead investors put these market people in charge of these companies and ultimately they win because they fail but can convince the government they need our tax dollars. Ultimately, they lead to higher profits because the tax payers make their profits. If an engineer was running the company, they might be winning and not need our tax dollars but their profits are still lower because they don't get our tax dollars. All these subsidies encourage crony capitalism and we get stuck doing this over and over. Just one more bale out for these massive companies. Then Elizabeth Warren, who voted to give these companies massive subsidies, complains these companies have massive profits. You are the one doing it!

Another example of how global trade and outsourcing to save ten cents ruined everything.

I am old enough to remember the days when people used to say that governments should not pick winners.

China will get its own duv and euv, its just a matter of time.

I bet the Intel CEO that made the terrible decision still got millions in bonuses, shares and compensation. They always do.

Fascinating insight into the world of semiconductors and how geopolitical dynamics have shaped its advancement. I've also found the role of EUV technology quite intriguing, bringing a new dimension to our understanding of technological progress. The quantum leap from a small Dutch company to a major players against Japanese companies is an inspiring journey. It’s disappointing that US companies did not capitalize on their investments, affecting their dominance in the market.

8:28 “Government funding” = taxpayers’ money. 😢

Cryptomining and AI hype bubbles have driven GPU chip demand has been primarily responsible for nvidias growth

Why..... Lobbyist were able to sway American politicians (Both parties) to abandoned American manufacturing for Cheaper Taiwanese labor.

I'm honestly glad that no single country has the means to make high-end chips. Hopefully that interdependence keeps us from killing each other

Many large American companies are too short term focus. Boeing is a perfect example.

Nice story bro framing it like the US made it all possible but it was Dutch ingenuity and persistence that made it work. It’s technology but how these machines work and how they even figured it out is closer to magic than anything else.

TSMC engineer like to say, lithography machines is just the oven, you can't make bread without bakers. the reason US lost the chip race (yes, China already produce more chip), isn't because it doesn't have the best oven, it because american don't want to be the baker, it a hot and boring job. Asian country are ahead because of their culture and willingness to work hard. when TSMC can't find "baker" in the US, the tried to bring them in from Asia, only to be stop by the unions, american unions are destorying US ability to "bake chip", it's that simple. people who talk about this like it a technology mistake, are just diverting attention away from the fact that american rather be youtuber than engineers...

2:40 I once overheard a process engineer complaining about unreasonable tolerances. One of the chip's layers was to be only 24 molecules thick, and the layer thickness should have no more than 3% variation.

2:00 How can a layer be polished to have a thickness of less than one atom? The video goes on later to correctly state that this is impossible at 2:47

*Still never buying intel stock.*

ASML controls the world. Proud to be Dutch.

Just the fact that this technology exists blows my mind. I can’t even wrap my head around what goes into this.

Applied Materials was the leader in chip making machines un til ASML. What happened on their end to lose out?

Why? Corporate greed. Not that hard to realize off shore labor and business tax breaks and lower environmental protection all lead to companies moving operations.

I’m working in Intels Fab34 in Ireland and it’s installing 16 EUVs. They are beasts.

Transistor processors have almost reached the limit of performance and power consumption. And most developers offer comparable technologies. This whole race for chips is very reminiscent of another bubble. A huge number of factories are being built and at the same time a huge number of export restrictions are being introduced. In the next couple of years this will lead first to an oversupply, and then to a big crisis in the industry. The future belongs to photonic and quantum processors, they are already a reality. And for hybrid processors, analog + transistor for AI needs.

When did 'chip making' become one word?

We are in between the very small and very large.

Here for the unnecessarily sick bass drop at 5:28

The main reason U$A will never able to compete in Chip manufacturing. All the country that do well in chip manufacturing , has Confucianism culture. For chip manufacturing, a high level of discipline is the key, and most Americans today don't possess it. They call it “forced labour" Taiwanese media reported on August 2 that TSMC claimed the production holdup at its Arizona facility was caused by a shortage of trained American labour and that they had sent staff from Taiwan to assist with the factory's development. Labour union officials in Arizona, on the other hand, criticised TSMC for exploiting this as a justification to bring in "low-wage foreign labour."

The USA and Europe need chip making, not next to china. This is an eggs in one basket issue that is being fixed.

I wonder if it is because these companies are seeking to develop products that can be sold only to governments or the military.

An interesting summary of the industry, just skimed the surface I know but, still thankyou.

Intel will never beat TSMC. They maybe can beat Samsung because intel a pure semiconductor company. but not TSMC

All poor America is sad they didn't steal it

Fun fact. For a employee incentive the employees had the choice to get shares or money. Some of the people working there since a long time are millionaires due to the bonuses back in the day when the shares were really cheap.. Even the cleaning crew

Everyone's gangsta until quantum computer shows up

So the EU and US is going to sanction China EV cars because the China government has subsidized it to make it competitive low price....and Now the US government is begging Intel other chipmakers to take the government grant money to make semiconductors monoploy...I am so confused..

The narrator voice is way too soft in volume. The difference between his voice and those interviewed is way too big also.

Fantastic explanation of the technology.

It’s because the money and policymakers thought they were above blue collar work and could get by pushing the more physically demanding and less profitable sectors to others. They invested all their time and energy into office jobs for finance, software and e-commerce and ignored the plight of everyone else.

Sooo Intel did a Kodak

I still think all these shenanigans against China are just the correct incentive for them to drop gargantuan amounts of money to develop the technology to catch up and surpass ASML.

Real strategy takes balls because of the inherent risk, but the only thing that Corporate America can stomach is cost cutting and outsourcing (which isn't a strategy since it's the default every company turns to). While it guarantees next quarter's returns, it also guarantees the company's long-term decline.

this is incredible Journalism. condense and informative. Excited for Intels next advancement... with that being said, how would it affect parallel processing chips like gpus.

So those machines will be odsolete with the next gen material used like graphene or photonic chip. So who will suffer more with China shunning US's chips due to national security, decoupling, derisking or overcapacities if you actually know what it means?

The two methods to get beyond 12nm are EUV and "quadruple patterning" on iDUV. EUV is easy, but expensive; quadruple patterning is cheap, but very difficult. Intel always planned on getting EUV machines, they just went with quadruple patterning first, which made them fall behind. Now that TSMC and Samsung have to develop their own quadruple patterning, Intel is catching up.

Name of the documentary is misleading.

TSMC overtaking Intel in CPU process size wasn't because of EUV. The first TSMC manufacturing process which overtook Intel was "N7", which was slightly denser than Intel's most advanced process at the time (10nm). N7 doesn't use EUV, it only uses DUV (Deep UV, the predecessor of EUV). However TSMC extended their lead by adopting EUV for N7+ and N6, which were enhancements of the N7 process; while Intel stayed on DUV for "Intel 7", which is what they called their enhanced version of 10nm. Intel will use EUV for their next-gen "Intel 4" process, but TSMC is already making chips (Apple M3) on an "N3" process which is significantly denser. Intel 10nm was a complete disaster, which ended up taking about twice as long to develop as Intel planned, while also being much less reliable and more expensive than expected. Meanwhile, TSMC N7 progressed faster and performed better than most people expected, resulting in Intel competing against AMD's 3rd generation of processors built on N7 (Ryzen 5000-series) with 11th gen Core CPUs built on Intel's old 14nm process from 2014, with half as many cores and double the power usage, for most of 2021, until Intel finally got 12th gen CPUs built on the Intel 7 process out.

Outsourcing = profits = wealth over nation health

It's a bit too late for the US, as China start to make their own High end chips starting from 7nm in 2023, then 5 - 3nm by the end of this year and possibly 2 - 1nm by 2025. Huawei spends more than 50 billion on innovations annually and has started to build Fabs for the New photon chips which are diamond based, there will be multiple layers and enough space to put everything into one 3d chip while using SHDC laser instead of electricity would mean its' hundreds times faster than the current silicone based chips, also the diamond based chips can release heat much more efficiently.

YO, BLOOMBERG! I can't hear your announcer. Maybe get him a mic that compensates for that intensity dropoff he does at the end of sentences? (I don't know the technical term.) I haven't finished the video yet, but so far the phenomenon is at its worst at 3:25 where I had to replay that section five times, gave up and turned on UKposts's hit-and-miss captioning to find out that he said, "match government spending". This is not to overly or unfairly criticize this announcer's delivery. I just think that this voice is getting swallowed up in the audio mix.

AND just to add to the string of mistakes in the US, the comeback in the chip industry which uses an enormous amount of water is being built in a state "known for it huge amount of water as a natural resource" - Arizona.

Here in the US, CEO's are keenly focused on the current quarter's EPS target, NOT the long-term. Same with US politicians, they only focus on the current Election cycle. Further, a US politician that has ZERO business experience and has collected a government paycheck for +40 years should just stay out of the way.

If you are going to claim Americans did the foundation for asml euv machines, I like some acknowledgement for us inventing the ship and the wheel that got you to the country you stole from native Americans. All technology is based on what came before. That doesn't mean you can claim everything as your own. Others have added crucial bits you didn't.

Then all of a sudden a new tech comes along making the other outdated.

It is bully and NOT a competition when only one side(U$A) can change rules. Just like your classmate(U$A) use his family power coercing the stationary shop(ASML,Japan,..) NOT to sell stationary to you.

Chloe Grace Moretz working undercover as a journalist.

Intel started with a design that must never survive the 80th from last century., the 8086.Two argument instructions, a direction bit, baroque special functions attached to registers, too few registers, bizarre segmentation and to top it off, duplicate instructions to use the AX registers to save one (!) byte of instruction size. The stackoverflow boys infuriated me for shooting down the question"why is the i86 considered a bad design? ". The geniuses as AMD succeeded with the i86 as a base but the ingenuity was better spent elsewhere.

It's a classic story of greed and outsourcing manufacturing to pay lower wages where possible. Intel thought they were big enough to sit on top without innovating until smartphone technology swooped in and demonstrated better performance than Intel PCs. There would be no significant improvements on the CPUs each year, yet they'd charge the same if not more money for each successive product. They were probably making bank with such profit margins. Only after Apple stopped using Intel did that company begin to scramble. The writing was on the wall for Intel the last 10 years and they didn't care until Apple and TSMC built a coffin and said get in

In tech years. The US is 40 years behind at this stage.

what really irks me is that corporate America has no problem taking billions in government subsidies and handouts (your tax dollars), whilst stubbornly refusing to pay their fair share in taxes 🤷‍♂. It's socialism for the rich and brutal capitalism for everyone else.

It's amusing how there's very little talk of decisions, because fab production is exponentially more expensive than fabless design. DUV and esp. EUV machines are prohibitively expensive, so it's hard to justify the up front capital expense without a very solid business plan.

Interesting. I never knew EUV became prominent only relatively recently.

We are a capitalist nation, why are we funding businesses like a socialist nation. We should be consistent with companies as we are with people.

Nothing about the architecture that Intel uses compared to that mainly produced by TSMC. They portray it like if Intel just gets back to ASML everything will be alright. Intel uses x86 architecture which is inferior to ARM based architecture.

Was Intel's refusal to use the UV lithography machines the reason they were so slow to shrink transistor sizes? They were on 14nm forever. Their slowness to shrink the chips and lower the energy needed was perhaps the single biggest reason Apple developed their own silicon. But were those years of such slow progress all due to them not using UV lithography machines?

EUV machines represent a monumental leap in semiconductor manufacturing. What are the key technical milestones that had to be achieved to make EUV a viable solution?

I am an Indian and have worked with tons of managers from US. They just don't know the basics about their respective products and that's where they lack. Meetings ppts and conferences are not going to help if the people at helm don't know anything about the product

intel will not be a leader again because of conflict in interest, if you are nvidia, and intel is making their own AI chip, will you send them your blueprint to make your AI chip? if you are AMD, and intel is making their own CPU chip, will you send them your blueprine to make your CPU chip? intel cannot be a national champion in chip production because it will always be treated as an inhouse fab which only 2nd rate companies who doesn't have any fear of their design been seen would be willing to use it. it really no rocket sciece why intel failed to attract business. AMD spinning off it fab business decade ago was the right decision because that was the only way for the fab to survive, it need its neutrality, and globalfoundries is now trying to get investment from US to build plant in US. those are the real competitors. talking about intel "winning the race" just show how clueless Bloomberg is.