I think there's a very large variation in estimates. Even from BLS itself. Compare the accepted answer with the more recent BLS data from a 2017 NYT article which show near parity for the 2014-2024 projections between job opening and US degrees in CS:
While using BLS data, the above analysis was performed by Edward Lazowska, a professor of computer science at the University of Washington.
Furthermore, graduating with CS/IT degree is nowhere near a guarantee for a job as one might incorrectly infer from the graph above. For instance, in the same NYT source, we read that:
Unemployment rates for STEM majors may be low, but not all of those with undergraduate degrees end up in their field of study — only 13 percent in life sciences and 17 percent in physical sciences, according to a 2013 National Science Foundation survey. Computer science is the only STEM field where more than half of graduates are employed in their field.
This somewhat jibes with the EPI report (which I myself) found pretty questionable in its inferences as it uses exactly one year (2009) as sample, but the findings of which, as pertaining to CS/IT, were eschewed in the accepted answer:
In computer and information science and in engineering, U.S. colleges graduate 50 percent more students than are hired into those fields each year; of the computer science graduates not entering the IT workforce, 32 percent say it is because IT jobs are unavailable, and 53 percent say they found better job opportunities outside of IT occupations. These responses suggest that the supply of graduates is substantially larger than the demand for them in industry.
So deficit of jobs and surplus of graduates in IT/CS? Skills mismatch? Hmm...
And if I'm allowed some related thoughts, although the question was tagged with US, the situation that CS/IT graduates fare better than the STEM average in finding a job in their field of study doesn't seem to hold in other Anglophone countries; e.g. in Australia. So frankly any inferences based on US data of a global shortage of developers seems even more tenuous.
And frankly taking the linear extrapolation used in the accepted answer to forecast millions of unfulfilled IT jobs is ridiculous, as the EU has found out:
“In 2011, the European Union was faced with 300,000 unfilled vacancies in the ICT sector; if this trend is not checked, there could be as many as 900,000 unfilled vacancies by 2015,” said the official conclusions of an October 2013 summit in Brussels.
“This skills mismatch is detrimental to our economic and social policy objectives,” added the text, signed off by all of the EU's government leaders, who promised action.
But in reality, the number of unfulfilled potential jobs in the ICT sector remained more or less the same in the following years, at around a third of the expected figure.
In 2012, the figure was around 274,000. In 2015, Empirica said there were 373,000 open posts, “so the actual shortage is not as massive ... as feared two years ago”. One year later, Empirica said the skills gap figure was 270,000.
Even assuming these gaps are real, they're not necessarily increasing, but showing some hard-to-predict variation:
In its reports, Empirica offered estimates based on different scenarios, but it was not uncommon for commissioner Ansip's predecessor for the digital agenda, Neelie Kroes, to take the highest figure, saying that “Europe faces an ICT skills gap of nearly one million workers”, without mentioning a year or any caveats.
The researchers had also stressed that what they calculated was a “demand potential” or “job potential”, not actual expected vacancies.
“It should be seen as a (theoretical) figure describing the demand potential for new ICT jobs which ... could theoretically and additionally be created in Europe due to an e-skills demand likely to occur especially in the years closer to 2020,” the researchers wrote in 2014.
“Vacancies that cannot be filled year after year will go away – projects cannot be realised, tenders not submitted, innovations will simply not be made”, they added.
That's why vacancies don't simply accumulate (linearly).
And to conclude my skepticism by coming back to the US, a 2017 Forrester report, as summarized in TechRepublic said that the purported unfulfiled jobs didn't translate in a substantial compensation increase, the widely accepted measure of such gaps:
Even high-demand tech roles have seen limited wage increases, the report found. Applications developers, security specialists, and other coveted positions have seen annual average job growth rates exceeding 7% over the last five years, while professions related to management and analysis of tech systems have grown at CAGRs about 3%. However, the wages for all of these roles have only growth around 3%, which suggests an efficient labor market, the report noted.
Further, over the past four years, the number of students graduating with degrees and diplomas in computer science has grown faster than the number of new tech jobs. "This suggests that any tech talent shortage is getting better — not worse — from a training and development standpoint," the report stated.
The report goes on to say that any "skills gaps" involve experience in hot technologies do jour (like mobile apps nowadays) and/or less attractive work locations; geographical areas identified (via salary growth): Rhode Island, West Virginia, Montana, and Washington State.