Within a few weeks after the break-out of COVID-19, organizations and governments established the remote-working economy. This new working practice resulted in many hardware appliances, ranging from high resolution cameras, microphones, and screens to CPUs, being required to perform an upgrade. In one way or another, all these upgrades fueled a general semiconductor demand.
By the end of Q1/2020, the semiconductor capacities were outpaced and a general shortage occurred. Long before, new industry activities, such as mining crypto currencies, focused on high performance processors (GPU and CPU). Nvidia and AMD were the main beneficiaries of this unexpected use of their hard- and software. It became evident how much the entire business case ballooned when, in mid-2022, GPU prices corrected by more than 50% on the back of the digital currency winter.
Ever since then, the semiconductor market has entered into a steep correction phase. The industry’s key markets such as PCs, smartphones, servers, and game consoles, experienced a sharp decline in demand. As did Intel AMD, Qualcomm, and Nvidia whose chip sets depend on DRAM and NAND configurations, thus inexorably intertwined with each other.
There is now a whole glut of flash memory, GPUs, and CPUs! It is therefore no surprise that during the last quarter electronic giants’ profits declined by as much as 90%. The most prominent victims were Samsung, SK Hynix, and Micron. But was this oversupply forecastable? About a year ago, the industry watchers at TrendForce issued some concerns about the built-up inventories across various sub-sectors. Today, the same team says that it will require at least another two quarters to eliminate the overflow.
Yet, this does not mean that the shortage of semiconductors is over! While the demand for high-margin products is gone, 1st generation chips used for consumer hardware are still lacking. These chips are integrated into power switch boards and the automotive industry, and the shortage is not done anytime soon.
The way forward
Next generation products such as LPDDR5, DDR5, and HBM will drive new demand. Typically, these next generation components come along with next-generation servers, high-performance products, and mobile applications. Companies such as Samsung and SK Hynix are expected to bounce back better and quicker on a relative basis on the back of an increased demand for these dedicated memory transactions.
Yet, despite this positive industry trend, the outlook is relatively grim for the quarters ahead. Here is a short run down of an industry analysis:
- Intel expects its revenues to fall to $11 billion and potentially lower in the 1st and 2nd quarter of 2023, followed by a turn-around by the hope that it can fix its issues with the 10nm process.
- AMD is predicting flat revenues for Q1 and poor PC and gaming sales throughout the 2023 fiscal year.
- Qualcomm again blamed the fact that nobody is buying phones and there is probably more to come.
- TSMC forecasts a revenue drop; its first in four years.
The general industry assumption is that demand should return, in the best case scenario, in the 3rd quarter 2023. This implies that the industry sector has another downside of some 12% to 18% ahead. Yet, even with this additional wash-out, the industry sector is not expected to re-experience the same performance as during the pandemic—the fundamentals are just not the same.
What lies further ahead?
With the US-China trade war focusing now mostly on semiconductor production capacities, companies such as Samsung, TSMC, and others are building new facilities in the States and Europe.
There are some 20 fabs under construction, all in the western hemisphere. Despite poor quarters, the subsector companies are expected to spend between $32 billion and $36 billion on these projects in 2023. These new facilities will be ready starting in 2025 and onwards. So in other words, when the next semiconductor megatrend is expected to start.
Yet, the real question is who is going to benefit from these new facilities: a) consumers, or b) some specific companies? In our view, there will not be a specific over-supply of CPUs and GPUs; therefore, consumers will be required to pay a relative steep price for a high-performance product. On the other hand, there are a just a few companies able to make the new fabs available. One of these companies is called ASML, which we consider the mother of all IT companies, ASML is expected to be one of the key beneficiaries of the present build-up of new fabs.