3 ms·
> This is changing a bit, but not existentially. Clusters exist in all fields because they have major advantages. This is particularly true for startups but op
by kanox 8y ago
> This is changing a bit, but not existentially.
Clusters exist in all fields because they have major advantages. This is particularly true for startups but opportunity exists in other areas as well.
The software industry is still more spread out than others and companies headquartered in the Valley have major operations in many other places (including the likes of Kiev).
- sonnyblarney 8y agoYes, it definitely makes sense for companies HQ'd in the Valley to have 'other offices' in such places. This is the de-facto Canadian strategy, when the outgoing Ontario Minister of Economic Development was adamant that 'Canadian developers were top-notch, and go for 60 cents on the US dollar, and we intend to keep it that way' i.e. basically indicating that he wanted to keep Canadians (or at least Ontarians) poorer for the opportunity to work for Cisco et. al. It's definitely a rational strategy in a way, but it's also self defeating because it precludes a competitive options. Imagine trying to hire talent to Toronto when said talent can 2x their income by going to the US? (I understand cost of living is different, but aside from housing, stuff is cheaper in the US, and many young people especially have a hard time seeing past the dollar signs). This would be an optimal strategy for a country in difficult situation, trying to lever their top talent into the 'beginning' of an industrial landscape (i.e. Ukraine, who haven't had a proper economic footing, basically ever). But it's sad positioning for countries with already advanced economies supposedly trying to be competitive. Canada's situation next door to the US, with a flexible immigrant population, a shorter history and weaker cultural ties vis-a-vis the US (i.e. it's not a big jump from Toronto to Chicago as it is from Berlin to Chicago, in cultural terms) ... creates kind of a specific situation. The clustering analogy is very good, but it's also very hard to do. They are trying here in Montreal with AI, but I'm not so sure it will fare very well. Hopefully it will work out.