Memory Chip Shortage Clouds Hybrid Bonding Adoption Outlook, Dutch Chip Equipment Maker Besi Shares Nearly Halve From Peak

Stock News
2 hours ago

Netherlands-based semiconductor equipment supplier BE Semiconductor Industries N.V. (ASX: BESIY) once saw its share price climb steadily on optimistic expectations for its next-generation chip technology.

Now, however, a global memory chip shortage has prompted the market to question how quickly that technology will be adopted, and investor patience is wearing thin.

The stock has fallen 14% this week and is down 45% from its June peak.

Just one day earlier, Bank of America downgraded BE Semiconductor Industries N.V. (ASX: BESIY) to "neutral." After UBS cut its earnings forecast and downgraded the stock from "buy" to "sell," the share decline widened further.

That added to the selling pressure, making BE Semiconductor Industries N.V. (ASX: BESIY) the worst performer in the pan-European Stoxx 600 index between June and September.

As shrinking individual chips becomes increasingly difficult, the entire industry is looking for other ways to boost performance.

One approach is to connect multiple chips together or stack them on top of one another.

Achieving that requires specialized tools such as BE Semiconductor Industries N.V.'s hybrid bonding equipment.

As the world's leading maker of hybrid bonding equipment, BE Semiconductor Industries N.V.'s flagship technology can directly connect semiconductor components, improving computing performance within a smaller space.

Investors had bet that chipmakers would adopt the technology quickly, but the memory chip shortage and cost concerns could further delay that transition.

Barclays analyst Simon Coles said: "There are a lot of other factors at play here, and those factors are driven by a large number of variables. Unfortunately, none of these things are within Besi's control." He rates the stock "equal weight."

Besi Shares Underperform Peers on Adoption-Speed Concerns, Down 45% From June Peak

For investors, one worrying question is this: if memory chip architectures are actually becoming simpler rather than more complex, is BE Semiconductor Industries N.V.'s technology still relevant?

According to independent research firm SemiAnalysis, Nvidia's planned AI accelerator Rubin Ultra, due next year, will use fewer memory layers than the current generation, a stopgap measure to ship more AI chips amid constrained memory chip supply.

BE Semiconductor Industries N.V. had previously expected that every memory component used in Nvidia's AI accelerators would become increasingly complex, eventually pushing chipmakers to adopt the Dutch company's technology to stack multiple memory chips into a package thinner than a credit card.

Besi Valuation Multiples Have Fallen Sharply This Year

Currently, BE Semiconductor Industries N.V. shares trade at about 29 times next year's earnings, less than half the valuation peak of roughly four months ago.

Still, Barclays' Simon Coles said that may not be the main issue for BE Semiconductor Industries N.V. investors.

He said: "For the past three to four years, the core investment thesis for Besi has centered mainly on hybrid bonding. If hybrid bonding adoption momentum improves, the stock performs well. And if there is uncertainty about the outlook for hybrid bonding, the stock comes under pressure, just as has happened since this summer."

AMD is one of the early adopters of BE Semiconductor Industries N.V.'s technology, having used hybrid bonding in its Ryzen 7 processors as early as 2022.

But adoption among memory chipmakers has been much slower.

SemiAnalysis analyst Myron Xie said this is because making hybrid bonding truly work is not simply a matter of using BE Semiconductor Industries N.V.'s equipment.

The process also involves costly preparation steps to ensure the surfaces of the two chips are nearly perfectly flat.

Myron Xie said that for high-bandwidth memory (HBM), the cost of adopting hybrid bonding "may not be worth the incremental benefit."

There is a trade-off: if hybrid bonding is abandoned and the number of stacked silicon layers is reduced, the workload that a single memory unit can handle becomes more constrained.

Myron Xie said that while this is not ideal for AI computing, there are still some workarounds.

Workloads can be spread across multiple memory units, and a new generation of AI models is also being designed to use less memory.

Analysts who are bullish on BE Semiconductor Industries N.V. view the issue from a different angle.

Ideally, if memory chip supply were ample, AI models should run on larger-capacity memory modules rather than smaller ones.

Therefore, if memory chip supply eventually catches up with demand, which Bernstein analyst David Dai believes could happen in 2028, BE Semiconductor Industries N.V.'s hybrid bonding equipment should come back into use.

David Dai said: "The investment thesis still holds." He added that over the long term, hybrid bonding technology remains necessary.

David Dai also cited South Korean semiconductor giant SK Hynix as an example.

The company has not yet adopted the technology on a large scale, but at a closely watched conference in August, SK Hynix highlighted the advantages of hybrid bonding technology.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Most Discussed

  1. 1
     
     
     
     
  2. 2
     
     
     
     
  3. 3
     
     
     
     
  4. 4
     
     
     
     
  5. 5
     
     
     
     
  6. 6
     
     
     
     
  7. 7
     
     
     
     
  8. 8
     
     
     
     
  9. 9
     
     
     
     
  10. 10