The issues tech companies have with immigration are about costs. The argument that is about "talent" that the AeroFS blog post makes is
not about talent distinct from cost, its about AeroFS's preference to not bear the cost (including time, financial cost, and risk) of developing talent to a particular experience level and their preference to have that done for them, and the fact that they found someone for whom that was done in a way they liked and that -- except inconveniently situated with regard to immigration.
If we were honest that the issues in immigration (on both sides) are about costs (except for those that are about outright racism), then we could address them more effectively.
There is a fairly direct value to immigrants and to certain others (e.g., in the case of those with pre-arranged employment, as would have been the case for the worker AeroFS write about, to their employers) of admitting specific immigrants. There are also social costs and benefits of admitting immigrants. (This also applies to non-immigrant visas, includingthe H-1B, which are, in a sense, peripheral to the issue of immigration, though its mostly what the tech industry is focussing on when discussing "immigration", which is one of many disconnects between the tech industry and the wider society.)
The basic structure of our immigration system -- in terms of the various immigration major categories and the decisions about which of them are unlimited and which are limited by quotas -- represents a judgement about the balancing of the social costs and benefits. Without challenging that basic judgement, there are sensible reforms that could be made which make the system better for everyone. The simplest that I see is to eliminate all the hard quotas in the family-based immigrant visa categories, and simply decide, in each category, on a fee to be assessed for supernumerary (above the now-soft quota) immigrants. (Generally, this should follow the structure of the family preference categories, with the more distant relation categories having higher fees.)
Instead of changing the structure of employer-sponsored immigrant visa categories, and to address some of the non-immigrant economic visa categories (particularly H-1B), what I'd strongly suggest is creating a separate no-quota (that is, all fee-based) individual immigrant visa category for candidates who are neither barred from immigrating to the US nor qualified in any of the existing family preference categories, with the highest fees, and allow economic entrants for work that is neither seasonal nor short-term to use that (and eliminate the non-treaty based non-immigrant visa classifications that this provision replaces, particularly the H-1B), while keeping the existing quota-restricted employer-sponsored immigrant visa categories.
If a company wants to give an employee an payment (up-front or in arrears) to cover the fee as a supernumerary immigrant (in either a family preference class or the new open class), they could through a relationship governed under contract law, but they'd get no special position with respect to the employee under immigration law.
This would address any "talent" issue that isn't about cost that the tech industry has with H-1B limits (or that industry in general has with the limits in employer-sponsored immigrant visa categories), address those issues that are about costs by providing a direct mechanism by which the social cost of the cost savings to industry are internalized in the transaction, address the problems with hard quotas that cause real problems in the family preference categories which are both drivers of illegal immigration and undermine the entire purpose of family-centered immigration policy, and provide funds to address the social costs associated with the overall level of immigration.