5 ms·

They are also doubling down on fab tech to catch up with TSMC. Markets working as intended.

• mettamage 5y ago
  Well... I find that a bit of a stretch. I'd rather say we happen to be lucky. What if TSMC was a company that was about as good as Intel on specs and price. Would the market be "working" then? A new player might come along. That new player would need to have 20 billion dollars of money to play along though.
  • netcan 5y ago
    I don't disagree in the abstract. Massive entry costs and all sorts of structures can and do obstruct the "as intended" mechanics a lot of the time. It's a struggle to term the revenue dynamics of an Alphabet, FB or JPM as "markets" at all. Chips though... chips are a market and it is working as a market. IMO chips are a rare example of Real economics in the modern economy, as opposed to the intangible-only economy that used to be mostly banking. A notable feature of the chip market is the persistent demand, the ability to demand/consume more computing than chip manufacturers can produce. Compare to cars, say 100 years ago. Most people didn't have one yet. Demand could keep up with supply, markets grow fast, and also make consistent efficiency gains,. Eventually though, the market saturates. People have cars and just need periodic replacements. The market isn't growing. People still want lower prices, or shinier cars. If they get lower prices, the market will shrink. Efficiency gains in mature markets can degrow a market, if demand is saturated. If car factories becomes twice as efficient, we'll probably have fewer of them. Our demand is not that flexible. Same thing happened with smartphones and laptops, to a degree. They do what they do well and we only need one each. In order to have a learning curve anything like Moore's law, the chip market has to grow every year. That requires a lot of demand, to offset all the efficiency gains. I don't think a lot of markets have the demand potential to support a Moore's law. In this scenario, market's working pretty well.
  • AussieWog93 5y ago
    >That new player would need to have 20 billion dollars of money to play along though. From what I understand, quite a few national governments are at least looking into setting up their own local chip plants since semiconductors have become a critical industry. On that scale, $20b is not a huge speed hump.
    • VortexDream 5y ago
      All of these projects AFAIK involve enticing a company like TSMC to build new plants, not building their own competitor in the market. I don't think there's any appetite in Europe to invest tens of billions in building their own chip industry.
      • ddalex 5y ago
        > appetite in Europe to invest tens of billions in building their own chip industry. Why would they. They build the machines that create the chips. TSMC would merely manage the chip building orders and the consumables. In case of national emergency I doubt that any government would have qualms about nationalising the factories.
  • roenxi 5y ago
    > That new player would need to have 20 billion dollars of money to play along though. In 2020 Uber's net income rose to losing only 7 about billion dollars. And they are competing for a market far less interesting and defensible than advanced semiconductors. Competition would arise.