Just two weeks ago, we brought an industry-wide event forward until last year: NVIDIA had booked production with TSMC for its RTX 40, where the news was doing. And it is that those of Huang had to pay in advance a stratospheric amount for the Taiwanese to access the said reserve, where what seemed new now is a trend because the industry suffers from a blockage of production capacity: the prepayment reaches the chips.
There are no chips for everyone, not the pace that manufacturers demand of foundries. There is no industry that does not want this piece of silicon that positions it better in the market, automotive, IoT, audiovisual, science, laboratories and let’s not talk about users. A year has passed and everything remains the same, the improvements are slight and the relevant improvements will take another year to arrive, so partners from Intel, TSMC and Samsung don’t want to be left behind in the meantime. The solution? Financing with volume reserve right, or in other words …
Locking in chip production capacity
What was once only a signal to strike a deal is no longer so today. Intel and Samsung haven’t seen a documentation leak on this, but TSMC was not so lucky. How to create billion dollar FOBs in record time without going through a bank and in the midst of a crisis? Advance the technical samples of the wafers with the last tip node, then leave the production that you can offer on a plateau and thus request upfront payments for the chips that were not created to fund your new factories that will create them.
Without interest, without intermediaries and only with the commitment to fulfill what is expected, it has thus concluded three TSMC agreements with the three main players in the industry which are not Intel: NVIDIA, AMD and Apple. In the first quarter alone, TSMC reportedly saw its prepaid revenue skyrocket to new heights with 150 billion dollars which were finally filed by the three players in the third quarter of last year.
144% growth in this type of financing
In tasa de crecimiento trimestral (y dada la crisis del sector) el crecimiento de esta modalidad ha pasado a primer plano con un 144% de mejora, lo cual ha supuesto record histórico de la marca donde además will be hoped that in ingresos crezca a 26% , almost nothing.
The nodes are not just focusing on the new 5nm LP and HP, but the latter which will see the company’s FinFET technology: N3 or 3 nm. We don’t know how much went to one node and another, but what is certain is that the FABs will work with the latest scanners from ASML for high EUV, which should already be more or less reserved, as long as Intel leaves something for someone.
The production stall is total, the maximum performance has been reached by Fab and the improvements in these barely reach 20% being optimistic. There’s a bottleneck that money can’t fix right away, so the Big Three turned to the one that guaranteed them the solvency of wafers and chips. Will TSMC really be able to comply with what has been signed? In a few months, we will see the first results of this blocking of chip production capacities.
