On August 27, 2026, the Bureau of Labor Statistics published its 2025 to 2035 employment projections. Data scientists grow 35%. Computer programmers decline 7%. Both numbers come from the same model, and the line between them is a staffing decision that most companies are making right now without realizing it.
Every two years the Bureau of Labor Statistics publishes a ten-year employment projection. It is the least glamorous document in American labor economics and one of the few that nobody is selling anything with.
The 2025 to 2035 edition, released on August 27, 2026, is different from its predecessors in one specific way. It is the first decade-long federal projection whose base year already contains widespread enterprise AI adoption. Previous vintages projected forward from a world where AI was a line item. This one projects forward from a world where it is in the workflow, and the model has opinions about what that does to technology jobs.
The headline is modest. Total employment is projected to grow 3.5%, adding 5.9 million jobs and rising from 170.3 million to 176.2 million. BLS notes that this is considerably slower than the 10.9% growth recorded over the previous decade.
Computer and mathematical occupations do better than that, ranking as the fifth-fastest-growing major occupational group at 7.3%. BLS attributes it to the continued proliferation of digital tools and AI solutions across the economy.
Under that group average, the picture is no longer a single story.
What Do the 2025 to 2035 Projections Say About Technology Roles?
They split the technology workforce in two. Against a 3.5% baseline for all occupations, data scientists are projected to grow by 35% and add 95,400 jobs; information security analysts by 21% and add 40,600 jobs; and software developers by 10% and add 174,700 jobs. In the same decade, computer support specialists decline by 3%, network and computer systems administrators by 4%, and computer programmers by 7%.
That is not a rising tide. It is a sorting.
Two more figures fill in the middle. Computer and information research scientists grow 22%, adding 8,400 jobs from a small base of 38,600. Computer systems analysts are projected to grow 8%, adding 42,900 jobs.
And one number deserves to be pulled out of the group it usually hides in. BLS reports software developers, quality assurance analysts, and testers together as a single occupational page, projecting 10% growth for the combined group. Broken apart, software developers grow 10% while quality assurance analysts and testers grow 6%, adding only 10,600 jobs. The headline number that gets quoted reflects a real divergence between the people who write software and the people who verify it, which is worth noting given how much the current verification burden is growing rather than shrinking.
Why Does BLS Expect Three Computer Occupations to Decline?
Because it expects the tasks inside them to be automated, and it says so in its own words rather than leaving it to interpretation.
For computer programmers, the projection is that computer programming work will continue to be automated, helping programmers become more efficient in some tasks and save time and money. Many companies are expected to leverage technologies, including artificial intelligence, to automate repetitive programming tasks. BLS adds that some higher-skilled programming tasks will likely shift to other workers, such as software developers.
On computer user support specialists: employment is projected to decline as organizations continue to implement automated tools, such as chatbots, for troubleshooting. BLS then makes a point that most automation commentary skips, noting this use of automation may free up some support specialists to handle more complex cases and troubleshooting that require attention, but fewer are expected to be needed overall.
For network and computer systems administrators, the causes are more varied. BLS cites tasks increasingly being done by software developers focused on DevOps, tasks being outsourced to companies providing network-as-a-service, and administrators increasingly automating routine work themselves.
Read those three together, and a pattern falls out that matters more than any single percentage. In all three cases, BLS is not projecting that the work disappears. It is projected that the work will be absorbed, either by automation or by an adjacent role with a broader remit. The programmer's tasks move to the developer. The support ticket moves to a chatbot, and what is left of the support role gets harder. The administrator's routine work gets scripted by the administrator.
Absorption is a different planning problem from elimination, and it is the one most workforce plans are not built for.
What Do the Growing Roles Have in Common?
They sit where judgment has to be applied to something an automated system produced.
BLS is explicit about data scientists, projecting that growth stems from increased demand for data-driven decisions, the growing volume of available data, and firms' continued integration of AI-based systems into their workflows. Its phrasing of the role itself is the useful part: data scientists will be needed to help businesses leverage AI and other technologies to enhance their processes, make informed decisions, design and develop new products, and market their products more effectively.
Regarding information security analysts, BLS points to attack frequency and AI directly, noting that increased use of artificial intelligence and the rise of e-commerce have increased the need for enhanced security, contributing to the projected employment growth for these workers over the decade.
For software developers, the sentence that does not appear in the news release but does appear on the occupational page is: "Demand is projected to be strong due to the continued expansion of software development for artificial intelligence, the Internet of Things, robotics, and other automation applications."
Here is the comparison that most coverage of this release will miss. Information security analysts are projected to grow twice as fast as software developers over the same decade. If your 2027 hiring plan treats security as a function you add after the engineering team is built, the federal government's own model disagrees with your sequencing.
What Does This Actually Change in a 2027 Headcount Plan?
Three things, and none of them is "hire fewer people."
The roles you cut and the roles you need are not interchangeable. A support specialist whose ticket queue got absorbed by a chatbot does not become a data scientist because a spreadsheet moved them there. The decline projections and the growth projections describe different labor pools, and the retraining path between them is a real program with a real budget, not a reassignment. We covered what it costs in preparing your workforce for AI, and the honest truth is that most organizations underfund it by an order of magnitude.
The senior end is where growth is concentrated, and it was already the tightest part of the market. Data scientists, security analysts, and developers building AI systems are not entry-level hires, and the pipeline that used to produce them has been narrowing for four years. We wrote about that specific squeeze in the hidden cost of cutting junior developers. A ten-year projection that concentrates demand at the experienced end, in a market that has been under-producing experienced people, is not a comfortable combination.
Verification capacity is growing more slowly than production capacity. The 10% versus 6% split between developers and quality assurance analysts is the projection's quietest finding. Whatever your view on how much code agents will write, the people who confirm it works are projected to grow at a little over half the rate of the people who write it. That gap has to close somewhere, and in most organizations it closes by loading verification onto senior engineers who were hired to build.
If you are staffing against this, our guide to AI staffing models covers engagement structures, and our AI talent solutions explain how we build teams for this exact shape of demand.
How Much Weight Should a Ten-Year Projection Carry?
Less than a measurement and more than a forecast from someone selling something.
These are projections, not observations. BLS builds them from a model of industry output, productivity, and staffing patterns, and the model is revised every two years precisely because the world moves. The previous vintage, covering 2024 to 2034, produced different numbers for several of these same occupations. Anyone presenting a BLS projection as a fact about 2035 is overselling it.
Two things make this release worth more of your attention than that caveat might suggest.
The first is who produced it. BLS sells nothing, sponsors nothing, and has no product that benefits from its own findings. That is unusual in this subject area. Almost every widely circulated statistic about AI and employment comes from a company with a commercial interest in the conclusion, and we would include our own published observations about nearshore engineering demand in that category. A federal statistical agency that publishes a methodology and revises it publicly every two years is a genuinely different class of evidence.
The second is the direction of agreement. The projection points in the same direction as the monthly data, the seniority mix in job postings, and what we see in our own hiring pipeline. When a ten-year model, a monthly series, and a client-side observation all point in the same direction, that direction is the finding. The decimal places are not.
Use it as a shape, not a number. The shape says that inside technology, AI is not reducing headcount so much as redistributing it toward roles that exercise judgment over systems, and that the roles absorbing routine work are the ones being absorbed.
Common Questions About the 2025 to 2035 BLS Technology Projections
Which technology jobs are projected to decline by 2035?
Three computer occupations are projected to decline: computer support specialists by 3%, network and computer systems administrators by 4%, and computer programmers by 7%. BLS attributes the declines to automation of repetitive tasks, chatbot-based troubleshooting, DevOps absorption, and networks-as-a-service outsourcing, rather than to a drop in the underlying work itself.
Which technology jobs are growing fastest?
Data scientists lead at 35% growth and 95,400 added jobs from 2025 to 2035, followed by computer and information research scientists at 22% and information security analysts at 21%. Software developers grow 10%, adding 174,700 jobs. All are well above the 3.5% projected for all occupations.
Are software developers still a safe hire?
By this projection, yes, but with a caveat. Software developers are projected to grow 10% and add 174,700 jobs, roughly three times the all-occupations baseline. Information security analysts are projected to grow at twice the rate. Treating security hiring as something that follows engineering hiring runs against the federal model's own ordering.
Why do developers and QA testers show different numbers?
BLS reports them on one occupational page with a combined 10% projection, but the underlying data separates them. Software developers grow 10% while quality assurance analysts and testers grow 6%, adding 10,600 jobs. Verification capacity is projected to expand at roughly half the rate of production capacity over the decade.
How reliable are BLS ten-year projections?
They are model outputs revised every two years, not measurements, and the prior 2024 to 2034 vintage produced different figures for several of these occupations. Their value is that BLS has no commercial interest in the results, publishes its methodology, and publicly corrects itself. Use the direction and the relative ordering rather than the exact percentages.
If you are building a 2027 plan against a projection that splits your technology workforce in two, the useful conversation is about which roles you develop internally, which you hire, and which you staff through a partner. Talk to the Golabs team about how to sequence it.