Categories
Electronic Components

Process nodes and transistor density

There are regular news articles published claiming that the smallest ever process node has been produced. We hear all the time about how small chips are becoming. But how can we measure this progress and does size really matter?

Moore’s Law

The concept of Moore’s Law, loosely, is that the number of transistors in a microchip increases as the size decreases. Originally, when Gordon Moore observed this in 1965, it was thought that the number of transistors would double every two years, but this rapid rate has definitely slowed.

Even so, there is still a constant increase in the number of transistors that can fit on an IC. In 1971, 6 years after the advent of Moore’s Law, there were around 2.3 thousand transistors on a single chip. This sounds like a lot, but we can now fit hundreds of millions onto one.

Nowadays, as it probably always was, it is a race between manufacturers to produce the smallest, most advanced chips. And with the advancement of manufacturing technology, the stakes are higher than ever.

Process nodes

The main method of measuring electronic component progress now is through process nodes. This is the term used for the equipment used for semiconductor wafer production. It describes the minimum repeatable half-pitch (half the distance between two identical features on a chip) of a device. It seems, though, that even this node measurement is no longer accurately used, according to some sources.

Some recent node announcements come from big players in the industry, including Intel, Samsung and TSMC. Taiwan’s largest semiconductor company, TSMC, recently announced that it would be converting its 3nm process node into 1.4nm. Critics, however, were not sure how possible this would be.

Samsung also recently revealed its plans to start manufacturing 2nm process chips in 2025. Additionally, Intel is planning on producing 1.8nm chips in late 2024. Part of the process of developing smaller process nodes is changing the technology involved in production.

What is the measure of a chip?

The method of measuring chips by process nodes is not entirely accurate and can be quite ambiguous. Some people have suggested chip density within the chip would be a better indicator of advancement.

While companies compete to develop the smallest process, some companies are fitting more chips onto bigger nodes. To put it in perspective, Intel’s 7nm process has 237 million per millimetre squared. In comparison, TSMC’s 5nm chip has only 171 million per millimetre squared.

So, although certain chips may have a smaller process node, it doesn’t necessarily reflect how advanced the chip actually is. Intel often uses density to describe its chips, because that is much more beneficial to them.

It’s a process

The question is, should all chips be measured this way instead of in process nodes? If process nodes aren’t accurate to their original definition, the measurements don’t indicate of the highest power chips out there. This might be confusing to consumers when choosing a manufacturer.

It will become increasingly difficult to measure in process nodes as chips get increasingly smaller. Many manufacturers are already making plans for when they begin to measure in Angstrom rather than nanometres. If the changeover from one measurement type to another was not confusing enough, if the measurement method is inaccurate, it may get very complicated.

Apparently, though, transistor count can be just as inaccurate because there is no standard way of counting them. The number of transistors on a single chip design can vary by 33-37% which is quite substantial.

The final node

Unfortunately, there’s no definitive answer on how to measure the advancement of chips anymore. Moore’s Law is far from dead, but is very much up to interpretation these days. Those purchasing or sourcing chips will have to have their wits about them.

For those sourcing chips, contact Lantek. We can source day-to-day or hard-to-find components with ease and can guarantee our customers the best price. Get in touch via sales@lantekcorp.com or call us on 1-973-579-8100 

Categories
Electronic Components

Price hikes in the electronics industry

More price increases announced

Chip prices will continue to increase, despite some component lead times improving. This is due to inflation, labour shortages, and scarcity of raw materials, among other things.

Intel was the latest company to announce price increases, which it will supposedly introduce at the end of this year. It joins firms including TSMC, Samsung, and Texas Instruments in raising the cost of its products.

As has become very clear, the pandemic contributed to supply shortages the world over. However, there have also been issues with labour shortages, material sourcing and the increasing costs of everything.

Reverse psychology?

Processors are increasing in price at Intel and other companies. It has been suggested that this actually may be due to oversupply. If the cost of the components is increased vendors are more likely to buy the stock before it occurs. As they stock up, Intel’s supply levels will decrease. This may lead to shortages in the long-term.

These increases are due to be introduced at the end of 2022, but people are suspicious it may happen sooner. If prices are instead increased in autumn, they can be discounted for events like Black Friday and Christmas.

War and price

Inflation is causing the price of materials to increase also, which inevitably would be passed down the supply chain. The price of raw materials was always going to increase over time, but the conflict in Ukraine has exacerbated this. Gases like neon, which is used in semiconductor production, is almost wholly (70%) sourced by Ukraine. Similarly, 40% of krypton gas is also from Ukraine, which is in conflict with Russia.

Aside from these materials, the price of lithium, cobalt and nickel, used for EV batteries, is also rising. The EV industry already had price hikes when the pandemic began, when the chip shortage took its toll. Now, following the 15% increase in 2021, automakers are facing another potential price increase.

Expansion

One of the largest players in the industry, TSMC, announced its price increases would take place in 2023. Despite not being as severe as first speculated, the 6% price increase will be enough that customers will notice.

Aside from the cost of raw materials, electricity and labour expenses, TSMC is also expanding. To fund this expansion it is increasing the price of fabrication.

Could we have stopped it?

Years before the pandemic, as far back as 2017, there were signs that a shortage was on its way. New technologies were mounting and other geopolitical difficulties were afoot. Even then, the best way to avoid this would have been to redesign the tech and improve the fabrication process. This would have been a time-consuming and expensive process, and whenever it happened it would result in delays and losses.

Conclusion

The amalgamation of all these factors will lead to lasting price increases for electronic components. Even if these prices are discounted in peak times like Black Friday or Christmas, suppliers will still have to deal with inflation and material shortages.

The expansion plans of some of the industry’s big players, and the cost of the tech to sustain them will also lead to price increases. How long the effects of these will last, we’ll have to wait and see.

Categories
Electronic Components

What is fabless production?

What is fabless production?

A fab is short for ‘fabrication’, which is a facility that produces electronic components. When it comes to fabless production, it refers to when companies outsource their manufacturing. The development of fabless production is a pretty recent development, but one that has flourished since its conception.

How did it come about?

Fabless production didn’t exist until the 80s, when surplus stock led to IDMs offering outsourced services to smaller firms. In the same decade the first dedicated semiconductor foundry, TSMC, was founded. It is still one of the biggest foundries in operation to this day.

In the following years many smaller companies could enter into the market as they outsourced manufacturing. More manufacturers, each with different specialities, also came to the fore.

Advantages

One of the original reasons it became so popular was due to the cost reduction it provided businesses. With the actual semiconductors being manufactured elsewhere, companies saved money on labour and space.

With production outsourced, companies also had the ability to focus more on research and development. No doubt this gave way to many advancements in semiconductor technology that wouldn’t have been possible otherwise.

Having a choice of which manufacturers to work with is beneficial too. Depending on your requirements you can choose someone who best suit your needs.

Disadvantages

When you outsource production, you are putting part of your business under someone else’s control, which can be risky. There could be a higher chance of defects if manufacture isn’t being directly overseen.

It also means that, in terms of quantity of product and price of production, you don’t have total control. If a manufacturer decides to change the quantity they produce or the price, customers are limited to their options. They either have to accept the changes, or search for an alternative which, in a fast-paced market, would be risky.

Conclusion/Disclaimer

The fabless business model, as it is known, will probably continue long into the future. TSMC’s continued profit, among other companies, is a key indicator of its success. And with big names like Apple, Qualcomm and Nvidia working fabless, it would be safe to say it’s popular.

That’s not to say that an integrated business model, with every stage of production occurring in-house, is a bad choice either. There are many just as successful IDMs like Samsung and Texas Instruments.

For a ‘fab-ulous’ stock of both foundry and IDM components, check out Lantek. We specialise in obsolete, day to day and hard to find electronic components. Send us your enquiry at sales@lantekcorp.com, or use the rapid enquiry form on our website.

This blog post is not an endorsement of any particular business model, and is purely for informational purposes.

Categories
component shortage Electronic Components

Global chip shortage to impact electronic retailers holiday season

The holiday season usually marks the start of an electronics sales boon for retailers. Consumers buy more electronics in the lead up to Christmas than at any other time of the year. This year, however, things are different.

This holiday season, the global chip shortage is set to impact electronic retailers, with shortages of popular products like games consoles, graphics cards, smartphones, laptops and tablets likely to persist through to 2022.

Due to problems buying stock, most retailers are bracing themselves for low Christmas electronics goods sales. The global chip shortage means fewer electronics goods are being made, so there is a long lead time from suppliers – some retailers are waiting several months for new stock, only for it to sell out within days.

Consumers should start holiday shopping now 

Chips are in critically short supply this year, which has reduced manufacturing output at many of the world’s biggest factories.

Companies like Samsung, Apple, Intel and AMD are experiencing problems getting the chips they need. Today, some chips have delays of over a year, and inventory supplies for chips are running low, putting pressure on supply chains.

All of this means there is a shortage of in-demand electronics goods, from games consoles to smartwatches. The message is simple – consumers should start holiday shopping now to ensure they can get hold of the electronics they want.

It is also crucial that consumers don’t take stock levels for granted. What’s in stock today might be out of stock tomorrow, and many retailers have lead times of several months for new stock. So, if you need it, you should buy it while you can.

Is the chip shortage being blown out of proportion? 

We are so used to next-day Amazon delivery and seeing shiny electronics on store shelves that chip shortages appear to be a fantasy.

However, the chip shortage is real – manufacturers are struggling to create enough chips, and suppliers can’t get hold of the inventory they need.

Categories
Electronic Components

Memory suppliers to benefit from strong demand and supplier shortages

While the downsides to electronic components shortages are well known, business is booming for smaller memory suppliers.

Sales of Samsung DRAM grew 26% in Q2 2021 without meaningful production capacity growth, and as supply/demand imbalances grow, memory suppliers like Samsung, Micron and others are turning to smaller suppliers to fill gaps.

As chip shortages continue, demand grows. Order books get filled off the page, creating longer lead times (up to 40-weeks) and extending standing orders. This is bad news for the end-product manufacturer but great news for suppliers, who see sales rise and bids increase to fuel record turnover and, in some cases, net profits.

The sector as a whole is booming, but no better example of taking the bull by the horns exists than Alliance Memory.  

Alliance Memory is a US-based 30-year-old DRAM manufacturer, billed as a legacy SRAM supplier and a leading domestic supplier of DRAM and flash memory. The company’s run rate in 2021 is double what it was in 2020.

In an interview with EPS News, Alliance Memory CEO David Bagby explains why: “we went out to customers struggling to get Samsung. Now we have maybe the best representation of DRAM and SRAM product of anybody out there.”

Memory upturn forecast to continue

IC Insights, the foremost authority on memory and chip demand, has predicted a new record high for memory demand in 2022.

Stronger DRAM pricing is expected to lift total memory revenue 23% in 2021 to $155.2 billion. The memory upturn is forecast to continue into 2022 to $180.4 billion, surpassing the all-time high of $163.3 billion set in 2018.

Demand for memory, including DRAM, SRAM, and flash, is being driven by economic recovery and the transition to a digital economy. Unlike other technological cycles, the current cycle of digitalization has weight behind it, fueled by innovations in data centers, 5G and space networks, AI, robotics and IoT.

Sequentially, the average price of DRAM rose 8% in the first quarter of 2021. Another increase of 18-23% in Q2 sent memory suppliers into a spin. Demand is outstripping supply, creating a perfect storm for continued price increases.

Price increases expected to continue until late 2022

The price of memory is more sensitive to other electronic components because supply is controlled by a few big players. Smaller memory suppliers fill in gaps in supply, but the big guns like Samsung and Micron rule the roost.

When demand outstrips supply at the big guns, prices explode. We’ve seen it several times before, such as the memory price increase of 2018. Prices fell again in 2019, recovered a little in 2020, then soared again this year.

Memory is a commodity and companies are willing to pay big to get a hold of it. Bidding wars are not uncommon and 40-week lead times are normal today.

However, while the memory upturn is predicted to continue into 2022, Gartner says memory prices will dive at the end of the year, predicting that an “oversupply” of memory chips will develop as demand eases and supply increases.

Categories
Electronic Components Semiconductor

Semiconductor production capacity expected to hit records highs in 2021

As you probably know, we are in the middle of a global semiconductor shortage. Auto manufacturers are cutting jobs, brands are delaying the release of new products, and people are struggling to buy things like games consoles.

It’s a grim situation predicted to last a few years, but behind the scenes, semiconductor companies are preparing to come out of the chip shortage swinging.

In fact, it’s predicted that semiconductor production capacity will reach a record high in 2021 so long as additional production lines are completed. This is reliant on production lines coming online following investments made at the beginning of 2018.

According to industry forecasts, next year, another 10 production lines for 300mm silicon wafers will be added worldwide. These will contribute millions of semiconductors each year, helping to release some pressure on demand.  

IC Insights also provides the following forecasts for chips: “By 2024, the average annual growth rate of semiconductor production capacity will be 5.9%. Compared with the average annual increase rate (5.1%) of semiconductor production capacity in the past 5 years (2014 ~ 2019), the growth rate slightly increased.”

Record demand for chips

The semiconductor market is experiencing record demand across all sectors. Chip manufacturers are struggling to keep up, but they are investing in new production lines to meet predicted demand several years from now.

The latest report from IC Insights’ McClean Report says that the semiconductor market will shake off the Covid-19 pandemic with 13% growth in 2021.

Semiconductor unit shipments are expected to hit 1,135.3 billion in 2021, fuelled by chips that target connected devices, VR and AR, network and cloud computing systems, contactless payment systems, automotive electronics including autonomous systems and consumer electronics including smartphones.

As technology advances and the world becomes more digital and more connected, chip demand will increase ten-fold over the next few years.

Semiconductor manufacturers are struggling to keep up with demand now but there are signs of life as the IC Insights’ report demonstrates.

The world’s biggest chip companies, including TSMC, UMC, SMIC, Samsung, Micron and SK Hynix are going to play a leading role in how technologies roll out long into the future. There should be no doubt these companies will power our future.

What next for semiconductors?

The prices of semiconductors are expected to increase by 20% in 2021 due to a shortage in production capacity and higher silicon prices.

However, the future may not be silicon at all. Graphene chips are 100 times smaller than silicon chips and thousands of times faster. This technology is in its infancy but it’s showing great promise. We expect big things in the next decade.

We also expect the semiconductor shortage to persist until 2022. Shortages should lift beyond this as production capacity increases from new production lines. Chip makers will need to manage supply and demand better in the future. The current shortage is bad news for everyone. Thankfully, it won’t last forever. Of this we’re certain.