How Profitable are TSMC’s Nodes with Math
We did some math on the profitability of TSMC’s nodes.
We did some math on the profitability of TSMC’s nodes.
TSMC reported earnings yesterday, and the management team seemed very happy about things. Their numbers came in ahead of expectations, and their outlook seemed very positive. AI is very good […]
If TSMC raises its prices as high as we are hearing they intend, then many companies will have no choice but to step off the curve of Moore’s Law. Maybe having an alternative like Intel is not such a bad idea.
TSMC has enjoyed 20+ years of a suppressed currency. Rather than squander that windfall on far-flung acquisitions, it has invested in its own talent pool, leading to the big advantage it has today.
To stay competitive in leading edge semis fabs, companies need to generate roughly double the cost of a fab in revenue. By this math Intel is cutting it pretty closely. It also helps explains some of Samsung’s decisons.
Some back of the envelope math. Backing out Nvidia’s numbers from TSMC’s results point to a good quarter from Nvidia, possibly a very good one.
The numbers on data center processor revenue share are stark. Nvidia has been growing strongly for years and now dominates the league tables. The only question is really what is the new status quo for their share?
We looked at revenue and operating income per employee for the big semis companies, and since that was so much fun, we looked at another dozen companies. Broadcom and Apple are in a league of their own. It is good to have a software or licensing business.
Intel faces three sequential problems: fix manufacturing, start designing competitive products again, and win some customers for IFS. Then they have to keep all that going. Not impossible, just very challenging.
We did some math to show market share and profit share in the foundry business. This data maps to intuition, but the scale of TSMC’s dominance and Intel’s declining fortunes stand out.