user nl
Well-known member
Question for this audience:
since the start of EUV in HVM around 2018 TSMC has typically invested >35% of annual revenue on capex, see this graph:
TSMC has (conservatively) predicted a CAGR of 25% till 2029, let's assume this will go all the way to 2030. This means TSMC revenue in 2030 will be (1.25)^5*122 B$ = 372 B$
If we assume TSMC will only spend 30% capex (so quite conservative) in 2030 TSMC will spend 0.3 * 372 = 112 B$ in capex in 2030.
TSMC capex is kind of linked to ASML's revenue and if you look at the ratio (ASML-revenue-in US$) / TSMS-capex since the start of EUV litho in HVM around 2019 we see:
Assume ASML will have a similar ratio to TSMC capex in 2030 it means ASML revenue is predicted to be about 0.86*112= 96 US$ ~ 96/1.17 ~ 80 BEuro in 2030.
ASML still predicts their revenue in 2030 will be between 44-60 B Euro. Their upperbound is way below the TSMC capex predicted revenue.
Will ASML-litho tools become a bottle neck for filling all the new fabs of TSMC and all the memory producers (for HBM) till 2030?
Any ideas?
since the start of EUV in HVM around 2018 TSMC has typically invested >35% of annual revenue on capex, see this graph:
TSMC has (conservatively) predicted a CAGR of 25% till 2029, let's assume this will go all the way to 2030. This means TSMC revenue in 2030 will be (1.25)^5*122 B$ = 372 B$
If we assume TSMC will only spend 30% capex (so quite conservative) in 2030 TSMC will spend 0.3 * 372 = 112 B$ in capex in 2030.
TSMC capex is kind of linked to ASML's revenue and if you look at the ratio (ASML-revenue-in US$) / TSMS-capex since the start of EUV litho in HVM around 2019 we see:
Assume ASML will have a similar ratio to TSMC capex in 2030 it means ASML revenue is predicted to be about 0.86*112= 96 US$ ~ 96/1.17 ~ 80 BEuro in 2030.
ASML still predicts their revenue in 2030 will be between 44-60 B Euro. Their upperbound is way below the TSMC capex predicted revenue.
Will ASML-litho tools become a bottle neck for filling all the new fabs of TSMC and all the memory producers (for HBM) till 2030?
Any ideas?