The "digital divide" started as a story about who has broadband. By 2026 it has become a story about who has the literacy, the devices, the time, and the institutional permission to use the broadband for high-value purposes — and who is on the receiving end of someone else's algorithm. The framing matters because the policy tools each version implies are very different. Bipartisan Infrastructure Law broadband investment closes the first divide. It barely touches the second.
The Pew Research Center's annual internet-use surveys put U.S. broadband access in 2024 at roughly 97% of households, with smartphone ownership above 90% of adults. The headline access numbers have closed substantially since the 2010s. The 2024 divide that matters is what workers can do with the access — and the OECD's PIAAC adult-skills assessment finds that about one in three U.S. adults can perform only the simplest single-step digital tasks. That number has barely moved since the 2012 baseline. Access without skills is not access in any meaningful sense.
The workplace consequences are large and growing
The Brookings Institution's Digitalization and the American Workforce series, led by Mark Muro, has scored U.S. occupations on digital-skill intensity and tracked them over time. Their finding is that the share of high-digital occupations roughly doubled between 2002 and 2020, and high-digital jobs pay 50–80% more than low-digital jobs after controlling for education. The wage premium is real and persistent. The Federal Reserve Board's research on rural-urban earnings divergence has documented that the gap correlates strongly with the local concentration of high-digital employment.
Generative AI is a second-order accelerant. The most rigorous early studies — including the Brynjolfsson, Li, and Raymond paper in the Quarterly Journal of Economics (2025) on AI in customer service, and the GitHub Copilot productivity studies — find that AI tools deliver disproportionate productivity gains to lower-skill workers within high-digital occupations, which compresses inequality within these jobs while widening it between high- and low-digital occupations. The within-job leveling is real; the between-job widening is real; the second effect dominates the first.
What employers can actually move
The corporate-responsibility framing of the digital divide has often produced underwhelming results because the interventions chosen tend to be visible rather than effective. Donated laptops are visible. Cohort-based applied skills training tied to specific employer hiring pipelines is effective. The empirical gap between the two is enormous.
The most rigorously evaluated interventions in this space are sectoral training programs. MDRC's evaluations of Per Scholas (an intensive IT-skills program serving primarily Black and Latino adults without four-year degrees) found earnings gains of about $7,500 per year sustained at two and three years post-program. Year Up's randomized control trial, published in the American Economic Journal, found gains in a similar range. The common features of programs that worked: 12–24 weeks of intensive applied training, employer partners with committed hiring slots, wraparound support (transportation, childcare, stipends), and explicit job placement. The common features of programs that did not work: shorter, more generic, less employer-tied.
Microsoft and Cisco have both built corporate-philanthropy programs around the sectoral model — Microsoft Skills for Jobs, Cisco Networking Academy — that fund partner organizations rather than running training directly. The data is more mixed than the press releases suggest, but the directional choice is the right one. The lesson is that the highest-leverage corporate move is funding scaled sectoral programs and committing to hire from them, not running internal training that is too short and too generic to move outcomes.
The accommodation gap that often gets folded in
One real category that is often subsumed under "digital divide" is workplace accessibility for disabled workers. The WebAIM Million 2024 annual analysis found that 96% of the top one million web pages had detectable WCAG accessibility failures. The U.S. Department of Justice's 2024 rulemaking on ADA compliance for state and local government web content has begun pressuring private-sector practice, and the EU's European Accessibility Act, in force from June 2025, will accelerate the convergence. Employers serious about this should treat web and software accessibility as a basic engineering standard rather than as a CSR initiative. The NWLB Disability Inclusion at Work → pillar covers the policy frame in more depth.
The policy frame
The Bipartisan Infrastructure Law made roughly $65 billion available for broadband expansion through the BEAD (Broadband Equity, Access, and Deployment) program. The Inflation Reduction Act and the CHIPS and Science Act made additional dollars available for workforce development tied to advanced manufacturing and clean energy. Together, this is the largest single federal investment in digital infrastructure and workforce development in a generation. The GAO's 2024 implementation reviews flagged that workforce funds are flowing disproportionately to incumbent providers rather than to sectoral models with stronger evidence — which, if uncorrected, will produce a familiar pattern: large spend, modest outcomes.
The state-level pay transparency laws (NYC, California, Colorado, Washington, Illinois) interact with this in a useful way. As employers are required to publish wage ranges, the wage premium attached to high-digital roles becomes more legible to workers and to policy advocates, which strengthens the case for investment in the training pathways that lead to those roles.
The honest CSR agenda
Companies serious about the workplace dimension of the digital divide should do four things. First, fund the sectoral training programs with evidence behind them. Second, commit to hiring from those programs with named slots and reporting on hire rates and retention. Third, treat web and software accessibility as a basic engineering standard, not a charitable initiative. Fourth, support — rather than oppose — the broader public-policy investments in broadband, public-sector skills training, and pay transparency that complement private-sector effort.
The companies that do this well will end up with stronger talent pipelines, better-trained mid-career workers, and reputational advantages that matter in tight labor markets. The companies that limit themselves to laptop donations and press releases will continue to be invisible in the workforce data.
The "digital divide" in 2026 is a literacy problem and an algorithm problem, not an infrastructure problem. The companies whose CSR programs still center on donated devices are answering the question of fifteen years ago.
Updated May 21, 2026. This piece was substantively rewritten as part of NWLB's 2026 editorial refresh.
