I'm inclined to disagree to be honest. If a data scientist (here in the UK) might earn (say) £50k, then to supply them with MATLAB and a comprehensive set of toolboxes, you're talking at least 15% on top of their salary. Then you need to employ someone equally skilled to rewrite and productionise their work, instead of immediately deploying it... perhaps another 35% on top.

The time and organisational cost of having that extra step is punitive, too: it's a well known chasm that great R&D frequently doesn't make it into production, and this is exactly why.

So MATLAB would have to make an already competent data scientist dramatically faster.... like, 33% faster to break even, without counting the organisational cost. I'm very sceptical that's achievable, particularly since open-source tools have orders of magnitude more tutorials and examples out there that junior and senior developers alike can leverage.

Also, don't forget that while in a perfect world one can do maths like "if a developer is more effective then..." - but in the real world budgets are limited. Even if it worked out more cost effective in theory, the additional CapEx for that software can be crippling for bootstrapped/seed stage startups (it certainly was for mine!) or any department whose budget is restricted or where HR budget is accounted differently from other expenditures.