For all the attention paid to whether new technologies will eliminate jobs, we struggle to answer the flip side of that question: What new jobs are emerging?
Part of the challenge is that these roles defy traditional categories. Policymakers project workforce needs using strict taxonomies. They define industries with NAICS codes, categorize workers with SOC codes, and look to O*NET to dictate required skills.
Government statistics must prioritize reliability, and there is enormous value in that consistency. But rapid technological change rarely respects the normal way of doing business.
For the past year as part of the research for my upcoming book The New American Frontier: Job Training for the Next Technological Age, I have researched the "frontier economy"—the jobs critical and emerging technologies create. As I recently explored in a new piece for Brookings, traditional labor market data is struggling to keep pace with the speed of emerging industries like AI, quantum, and new energy technologies.
Frontier industries are hard to classify
Understanding where new jobs are emerging starts with a basic definitional problem: government classification systems were not built to quickly capture new industries.
In the NAICS system, every business maps to a single primary code based on its main revenue source. But frontier fields don’t fit neatly into a single bucket. Quantum technology is a prime example. Because there is no dedicated quantum code, companies in the sector are scattered across the economy, classified as computer hardware manufacturers, software publishers, precision instrument makers, or specialized engineering services. Meanwhile, tech giants conducting cutting-edge quantum research are simply classified under their parent company's broader retail or software codes.
As the breakdown below shows, looking for "quantum jobs" in public data requires navigating a maze of disconnected industries:
Even when a frontier industry does have a defined industry classification, like semiconductor manufacturing, the headline code may only tell a fraction of the true jobs story.
Expanding domestic chip production requires the construction of new fabs, installation of specialized machinery, and supply of industrial gases and chemicals. Much of this early hiring happens across the supply chain before a facility goes online.
In our analysis of Revelio Labs job postings from 2023 onward, including upstream supply-chain sectors nearly tripled the volume of semiconductor-related postings compared to the core manufacturing code alone. Preparing workers solely for jobs inside the headline industry would leave a big part of the actual employment ecosystem out of the picture.
Sometimes the worker exists before the occupation does
New technologies also produce combinations of work that do not yet have their own occupational category.
Consider the emerging role of a biomechatronics technician. Modern pharmaceutical and biopharmaceutical manufacturing facilities increasingly rely on sophisticated automated equipment. Keeping those facilities running requires workers who understand regulated manufacturing processes but can also troubleshoot mechanical systems, electronics, industrial controls, and automation.
That role is more specialized than general maintenance work. But today, there is no SOC code for a biomechatronics technician. Depending on the employer and dataset, these workers may instead appear as industrial machinery mechanics, mechanical engineers, bioengineers, or another existing occupation. Looking only at those categories makes it difficult to see that a distinct type of work is emerging.
A different approach to worker profiles can help make these new roles more visible.Rather than starting with an occupational code, we can identify workers from their actual employment histories and experience. Using Revelio Labs profiles, we find that the number of workers focused on maintaining machinery within pharmaceutical and biopharmaceutical manufacturing increased from roughly 26k in 2008 to 57k in 2025.The job appears to have real demand, but it is not yet showing up clearly in traditional sources.
Industry and educators are already responding. Companies in North Carolina’s biomanufacturing cluster are actively recruiting this exact hybrid skill set, and Wake Technical Community College has launched dedicated training specifically for biomechatronics technicians.
If we rely only on historical data to tell us what jobs to train for, we may lock out emerging fields key to economic competitiveness.
Sometimes the title doesn’t change at all
Sometimes a job might be changing dramatically without acquiring a new name.
Think about an electrician installing modern battery systems.The job posting asks for an electrician, and they still need the foundational knowledge, training, and licensing. But the technology this person deploys now demands skills that were not central to the role a decade ago, such as managing bidirectional power systems and battery infrastructure.
We can think of these as retooled jobs: existing occupations whose bundle of tasks and skills changes as new technologies are introduced.
If the job title stays stable, it can mask the big changes needed in education and training. Many electrical programs still focus heavily on traditional lighting and outlets, missing the fundamentals needed for solar and batteries. One possible reason is because data sources like O*NET are designed to reflect the "typical" electrician, rather than the worker at a firm on the technological frontier. This creates a structural lag, making it difficult to capture what a job is likely to look like in the future.
Different data answer different questions
It’s important to note that this is not an indictment of traditional labor market statistics.
In fact, much of the strength of these measurement tools comes from their stability, and the fact that they do not redefine occupations every time a new job title appears. Consistent classifications give us representative, comparable measures of employment across the economy and over long periods of time.
The tradeoff, though, is speed and flexibility.
New sources of labor market data can complement the existing public infrastructure by looking at real-time indicators, like job titles, career trajectories, and company-level hiring.
Sometimes those additional data sources provide an earlier signal when the market changes course. In 2025, the BLS projected a decade-long 8% decline for nuclear technicians. Yet on the ground, the landscape was shifting: as policy and market demand evolved, the commercialization of next-generation technologies like microreactors and small modular reactors accelerated. According to Revelio’s data, job postings for nuclear technicians rose 68% between 2024 and 2025. The BLS eventually caught up to the trend, revising its next annual forecast to project modest growth.
Doing better doesn’t require perfect models or foresight. It means combining the right tools for the right decisions, and recognizing where data have limits.
