Large shares of Americans are connected to the world of digital information while on the go via smartphones and other mobile devices. Explore the patterns and trends that have shaped the mobile revolution below.
The internet represents a fundamental shift in how Americans connect with one another, gather information and conduct their day-to-day lives. For more than 25 years, Pew Research Center has documented the internet’s growth and distribution in the United States. Explore the patterns of internet and home broadband adoption below.
They tell the story in a series of graphs:
Seeing how data centers have churned up local politics and spurred newcomers to run for office, some city administrators came together to seek help. The end result is a new toolkit developed by researchers at the University of Minnesota to help local government officials better understand and weigh the pros and cons of having Google, Meta or some other big tech company build a massive data center in their community.
What does the toolkit include?
The guide, which will be made available to the public Wednesday, includes sections on how data centers can impact energy and water usage, and how these impacts can be regulated at the local level to manage environmental impacts. It also includes guiding questions for city leaders to ask if they get an inquiry from big tech about a data center proposal as well as examples of how to address data centers in local code.
The Minnesota Star Tribune reports…
More than two-thirds of Minnesota voters worry that artificial intelligence could someday destroy humanity, according to a new Star Tribune/KARE 11 Minnesota Poll, an extraordinary fear that joins more familiar anxieties about jobs, politics and the future of American life.
The poll found that 68% of likely voters are very or somewhat concerned that advanced AI could bring about humanity’s destruction. Nearly a quarter are very concerned. And 71% favor government regulation of the industry, a rare point of broad agreement among voters otherwise divided over much of the country’s direction.
The findings capture a peculiar moment in the technology’s ascent: AI is becoming an ordinary feature of daily life, even as many Americans remain uncertain about where it is headed — or whether anyone can control it.
Schools, Health and Libraries Broadband Coalition (SHLB) takes a look at the economics of discontinuing (or reducing) E-Rate programming, which pays for a considerable amount of broadband connectivity in schools and libraries…
The Federal Communications Commission is considering eliminating or significantly reducing E Rate support ($2.6 billion in 2025) for broadband connectivity and related services provided to schools and libraries. While eliminating the program would reduce Universal Service Fund expenditures, it would also increase the effective cost of broadband services for public and private K-12 schools and libraries that increasingly rely on digital connectivity to support teaching, learning, school operations and security, cybersecurity, internal and external communications, and access to online educational and informational resources. Nearly every critical function in schools and libraries now relies on the network, including door locks and access controls, security cameras, bus scheduling and transportation systems, cafeteria
payment systems, payroll, student information systems, attendance and grading platforms, and communications with students, staff, and families. Because broadband has become an essential educational input rather than a discretionary expenditure, schools and libraries cannot simply replace connectivity with teachers, books, buildings, or other educational resources. Paying more for connectivity therefore leaves less money available for the other goods and services needed to provide education and library services.
We use Brattle’s BEYOND computable general equilibrium (“CGE”) model to estimate the economy-wide contribution of E-Rate. Our primary counterfactual eliminates the program while recognizing that broadband and internal network infrastructure remain necessary inputs. The results indicate that eliminating approximately $2.6 billion in annual E-Rate support would reduce real GDP by approximately $4.2 billion annually, real wages by $2.8 billion, and tax revenues by approximately $1.5 billion. Thus, each dollar of E-Rate support is associated with approximately $1.60 in GDP. The measured economic loss exceeds the amount of federal support removed because without E-Rate, schools and libraries must devote more of their own budgets to connectivity, leaving less available to support the overall level of education and library services. As school spending generates greater follow-on economic activity than
broadband spending, institutions purchase fewer non-broadband goods and services from other suppliers and use less labor, and those reductions propagate through household income, spending, and tax revenues.
Eliminating just support for connectivity infrastructure (Category 2), including routers, switches, Wi-Fi access points, cabling, managed internal broadband, and related infrastructure needed to distribute connectivity within schools and libraries, would also have substantial effects.
Eliminating Category 2 alone (approximately $1.2 billion in annual support) produces an estimated $2.5 billion decline in GDP, $1.6 billion reduction in real wages, and $0.9 billion reduction in tax revenues.
Finally, we find that even unsubstantiated concerns about waste, fraud, inefficient procurement, or otherwise unproductive expenditures do not materially change these
conclusions. Even under a deliberately conservative assumption that 2.5% of E-Rate support generates no intended economic benefit given that the improper payment level has already been found to be lower than 1.5%,1 the associated GDP loss is approximately $113.9 million, or
only about 2.7% of the GDP loss from eliminating E-Rate altogether. The results demonstrate that eliminating a productive program to prevent a comparatively small amount of ineffective spending can impose economic costs substantially greater than the inefficiency being addressed.
The FCC’s review should therefore distinguish between monetary savings and economic benefits.
Roberto Gallardo of Purdue University has a new study (An Evolving Digital Divide:
Measuring and Understanding Mobile-Dependent Households (MDH)) out looking at folks with mobile-only access to broadband…
Despite unprecedented broadband investment, the share of U.S. households relying solely on cellular data and mobile devices or mobile-dependent households (MDH) increased from 11.9 to 14.1 percent between 2019 and 2024. This report introduces the MDH concept, which captures
the “double whammy” of limited data and limited screen size, and examines its determinants before and after the COVID-19 pandemic using Census tract-level data. Results indicate that the share of the population with any disability, households with a lack of internet access, income, educational attainment, and homeownership are among the strongest factors associated with mobile dependence, with several associations shifting in strength post-pandemic. As communities implement BEAD-funded broadband projects, these findings suggest that connecting households is insufficient if that connectivity comes through the most constrained pathway available.
I’ve been looking at reports about computer vs phone vs tablet access for years. I am acutely reminded of the difference on vacation. I can do many things on my phone, and in a pinch, I could do my job but it’s much more difficult. Roberto looked at 5-Year American Community Survey (ACS) dataset ending in 2019 and in 2024 to view mobile-only access.
Here are some images of what it looks like
His conclusions…
To summarize, this report introduces the concept of mobile-dependent households experiencing the “double whammy” of relying solely on cellular data and mobile devices and examines the factors associated with this growing population before and after the COVID-19 pandemic. Recalling that the share of MDH increased from 11.9 to 14.1 percent between 2019 and 2024, a period of unprecedented federal broadband investment. This growth, combined with the expanding footprint of high-high clusters shown in the LISA maps, suggests that infrastructure deployment alone does not resolve the structural conditions pushing households toward mobile-only connectivity. This is
important to keep in mind as the Broadband Equity Access and Deployment (BEAD) investments across the country start to materialize and the future of the Digital Equity Act funds, which targets skills and devices is uncertain.Several post-pandemic shifts stand out: the metro-nonmetro gap widened as metro tracts gained a significant negative association with MDH by 2024; disability’s association with MDH strengthened substantially, reflecting the consequences of essential services moving to digital platforms that small screens and capped data plans cannot adequately support; the no-internet association also
strengthened, indicating that where fixed broadband is absent, the households that do come online are increasingly doing so through mobile; income’s association strengthened as well, consistent with fixed broadband becoming more of a premium good in the post-pandemic economy; and the
association with the share of the population aged 65 or older disappeared entirely, suggesting that older adults who entered the digital world during the pandemic did so disproportionately through smartphones rather than fixed broadband. Notably, blue-collar occupations, while still positively associated with MDH in both years, weakened in strength, possibly reflecting that broadband expansion reached more blue-collar communities post-pandemic even as afordability became a more dominant barrier.
And cutting to the chase…
In moving beyond the question of whether households are linked to how they are connected, this report reveals that a growing share of American households experience a second-class form of digital participation. As Napoli and Obar (2014) warned over a decade ago, treating mobile access as equivalent to fixed broadband risks rendering this growing population invisible. The MDH concept makes them visible.
Using Ookla Speedtest Intelligence® data, this report identifies the states that are currently delivering the minimum standard for fixed and satellite broadband speeds as established by the Federal Communications Commission (FCC) to the highest percentage of Speedtest users. It also singles out the states that need the most improvement when it comes to delivering the minimum standard for broadband to residents.
Here are a few of their top takeaways:
Minnesota ranks 25 for access to fixed broadband at of 100/20 Mbps with the following:
Minnesota ranks 5 for access to users achieving Starlink LEO broadband speeds of 100/20 Mbps with 55.0 percent.
Now a community survey conducted by an area resident has landed like a hand grenade in the middle of the fray — and may shed light on that question. The survey suggests broad opposition to the data center with 75.8% of respondents opposing it and 19.2% supporting it.
“Do residents of Pine Island city and nearby townships support the proposal to build a hyperscale data center in Pine Island. The majority answer is: No,” concludes the summary of the 73-page “Pine Island Sentiment Survey” authored by Susanne Blazek.
Blazek has a doctorate in social psychology with a minor in statistics from the University of Minnesota. More to the point, she has spent the last 15 years of her career conducting sentiment polls and large-scale public research, she says. Blazek makes no bones that she opposes the data center.
More on the survey…
One reason Blazek thinks the survey’s conclusions are valid is that the findings are in line with national polls, which show a strong backlash toward data centers. Surveys from organizations like Gallup, NBC News, and YouGov show roughly 65% to 70% of Americans opposing constructing a data center in their local area.
Blazek said the survey’s findings are based on 656 verified responses from people in the Pine Island city and surrounding areas. The survey reached 4,000 people through flyers and online messages for a 16% response rate, “which is considered OK.”
National Telecommunications and Information Administration is looking for feedback on their Internet Use Survey…
NTIA seeks feedback under the Paperwork Reduction Act (PRA) on a set of questions for the November 2027 edition of the U.S. Census Bureau’s Current Population Survey (CPS). This collection of questions is known as the NTIA internet Use Survey and is also referred to as the CPS Computer and internet Use Supplement. NTIA has sponsored eighteen such surveys since 1994.
You can send an open ending email. They also have some suggested questions…
The bureau also welcomes comment on any of the following questions. Where possible, cite question numbers and propose specific, tested language:
Is data from the NTIA Internet Use Survey important in 2027? Should the survey be continued?
Which NTIA Internet Use Survey questions should remain unchanged in 2027 to preserve useful longitudinal time series?
How should the survey questions or language be updated for technology use in 2027?
Which questions should be kept because they are used by researchers, policymakers, and/or industry? Please explain how the question(s) are used.
Which questions use outdated language or will not provide useful data in 2027? Should these questions be removed? If not, how should they be changed?
Internet connectivity and applications have changed substantially since the survey instrument was substantively rewritten a decade ago. For example, the role of the internet in Americans’ lives is often unnoticed, from smart appliances to single-player video games that require online connections. How (if at all) should questions on the NTIA Internet Use Survey be edited, removed, or added to address the evolution of these technologies and the words we use to describe them?
Commenters often suggest adding questions, but space for questions is limited. If your comment proposed additional questions, are there other questions that should be removed to make space? Or would some questions be better asked in a different venue?
Are some questions (or potential questions) particularly important for policymakers or industry? Please explain. For example, as a supplement to the CPS, the nation’s primary source of labor statistics, the survey might provide key insights on internet use and the workforce.
What questions (or potential questions) are of particular importance for enabling high-quality evaluation of NTIA’s broadband grant programs and other federal broadband programs? Are there particular gaps in existing program evaluation data that could be filled by adding certain questions to the NTIA Internet Use Survey?
In what ways can survey language be customized to support cross-country comparison?
Is there a need for more data around children’s use of the internet and digital devices? Are there particular areas that need more data, such as screen use in schools or parental perceptions of children’s tech use? Is the NTIA Internet Use Survey, a CPS supplement, the most appropriate place for this, or should that data be gathered in other ways?
Is there a need for more data around artificial intelligence (AI)? Are there particular areas that need more data, such as AI use at work or home, or attitudes towards AI? Is the NTIA Internet Use Survey, a CPS supplement, the most appropriate place for this, or should that data be gathered in other ways?
Pew Research Center reports...
Between scrolling, notifications, and messages, smartphones can be hard to put down. Just over half of U.S. adults say they spend too much time on their smartphone, according to a Pew Research Center survey from May and June 2026. About a third of adults say their smartphone use is about right, while just 3% say they spend too little time using these devices.
The Federal Reserve Bank of Minneapolis reports…
The number of businesses currently using AI shows there’s potentially many workers to be trained. But much of the training needed is likely to be basic because most businesses that use AI don’t appear to be doing very much with it.
According to the U.S. Census Bureau, 22 percent of U.S. businesses used AI as of the end of July 2026. That’s a 4 percentage point increase since the beginning of the year.
Around the Ninth District, the share of businesses currently using AI ranges from 14 percent in North Dakota to 22 percent in Montana (see figure).
It seems like there’s interest…
Business demand for AI training appears strong, though much of it appears to be from businesses that aren’t sure about how to use the technology.
“What we’ve seen so far is the [client] organization wants to equip their people to use AI, but they don’t necessarily have an enterprise-wide strategy yet,” said the University of Minnesota’s Jahnke. “So they’re wanting to equip their employees with the basic knowledge with particular emphasis on how to use it ethically and a basic understanding of how AI works.”
And it’s growing…
Demand for training in AI skills is likely to grow with businesses’ growing use of the technology.
Over the next six months, the share of U.S. businesses using AI is expected to increase by 3 percentage points compared with July, based on Census Bureau surveys.
Among most Ninth District states, the expected increase is about 2 percentage points. Montana and North Dakota expected a decrease due to a growing number who expressed doubt that they would still be using AI.
Despite the technology’s limited impact on employment so far, there is a simmering fear in the public that massive job losses are not far off. Experts disagree on whether this is realistic, and where they land depends on how much artificial intelligence they think can replace human intelligence, and how soon.
Ayush Chopra, who developed a simulation of AI impact on the labor market while a grad student at MIT, said he doesn’t expect a “job catastrophe.” The simulation found that, among states, South Dakota had the highest concentration of jobs exposed to AI disruption, such as those in finance. By “exposure” he means that those jobs contain more tasks where AI can enhance human labor, not replace it.
If AI can improve productivity, that’s not a bad thing for South Dakota given the labor shortage, according to Hultman. The state’s unemployment rate in June was 2 percent, the lowest in the nation.
The National Lifeline Association reports…
The National Lifeline Association (NaLA) is proud to release the results of our 2026 Annual Subscriber Survey. With over 83,000 responses across 20 detailed questions, this year’s data confirms a critical reality: for low-income Americans, the Lifeline program is not merely a discount—it is essential access infrastructure.
Why Lifeline Matters
As regulators and lawmakers evaluate the future of the Universal Service Fund, our survey data provides a clear picture of how Lifeline empowers households by bridging the digital divide.
Key Findings: How Subscribers Use Lifeline
Lifeline provides the connectivity necessary to improve life outcomes and access vital services:
- Healthcare Access: 66% of subscribers use their service for telehealth appointments (phone/video) and to order medication.
- Staying Connected: 81% rely on the service to maintain connections with family and friends.
- Employment: 37% utilize the service for work-related activities, including working from home, applying for jobs, and managing shifts.
- Education: 30% leverage the connection for educational purposes, such as online classes, training, and schoolwork for children.
The Reality of Affordability
Affordability remains the primary barrier to sustainable internet access.
- 96% of subscribers report that they cannot afford any monthly service fee.
- 85% of subscribers would lose service entirely without the Lifeline benefit, as they cannot pay the full, undiscounted price.
- For those without home internet, the monthly bill is the most frequently cited barrier—outranking both availability and awareness.
A Mobility-First Requirement
With 91% of subscribers preferring mobile wireless service with hotspot capability over fixed broadband, it is clear that mobility is a functional requirement, not a preference, for this population.
The Path Forward
The current benefit level supports incremental use but falls short of providing the sustained connectivity needed for full, consistent access to telehealth, remote work, and coursework. Sixty-nine percent (69%) of subscribers report limiting their data use every month to avoid exhausting their allotment.
There is an interesting scholarly article (Time poverty, waiting, and broadband adoption: Evidence from the US by gender and income) on the impact of broadband access on people’s discretionary time. In a nod to time-saving, I tried to think of the most efficient way to share the information and focus on broadband access while flagging other related topics in the article.
Here’s the promise of the research, in bullet points…
And a look at ROI (when return in time) from a government/citizen perspective…
Addressing waiting-based inefficiencies calls for targeted solutions, and broadband and digital technologies offer one. To this extent, tax return–free filing programs that integrate administrative data could save Americans an estimated 225 million hours annually, while online voter registration cuts administrative burdens and boosts turnout by reducing long heavy processes (Goolsbee, 2006; Herd & Moynihan, 2018). Similarly, remote work saves the average person 72 min of commuting each day, allowing individuals to reallocate time to leisure and caregiving (Aksoy et al., 2023).
Yet, these time-saving gains hinge on both the provision of digital services online and citizens being able to take advantage of those through reliable internet connections (Mossberger et al., 2022). Consequently, broadband operates as the foundational infrastructure enabling both the delivery and use of digital services. In other words, lacking broadband, citizens, businesses, and governments cannot effectively use online services while digital service provision becomes impractical.
Recognizing this, federal and state policymakers in the United States have invested significantly in digital infrastructure – most notably through the 2021 Infrastructure Investment and Jobs Act (IIJA) and the American Rescue Plan Act – to extend broadband to every household. These place-based strategies reflect a growing understanding that broadband is not just a technological solution, but a foundational element of equitable service delivery and local development. As such, we argue that, similarly to other place-based characteristics, broadband adoption can shape individual outcomes through institutions, resources, and community dynamics (Chyn & Katz, 2021)
And their hypotheses:
The GAO (Government Accountable Office) reports…
The IIJA and IRA provided billions in federal funding to agencies to distribute through mechanisms like grants and loans. Beginning in January 2025, executive orders directed agencies to pause this funding to review it for consistency with administration priorities. Grant recipients have raised concerns that delays in distributing these funds affect their ability to implement their projects.
GAO was asked to review the status of IIJA and IRA funding. This report describes selected agencies’ (1) IIJA and IRA budget authority, obligations, rescissions, and disbursements for fiscal years 2022 to 2025; (2) processes for reviewing IIJA and IRA funding in response to executive orders; and (3) IIJA and IRA funding review status.
To address these objectives, GAO selected agencies with among the highest amount of combined IIJA and IRA funding. Of these agencies, GAO selected EPA, Interior, NTIA, and DOT for this report. GAO will report on the Department of Energy and Department of Agriculture separately.
GAO analyzed the IIJA and IRA and financial data from agencies’ accounting systems and the agencies’ policies and guidance for reviewing IIJA and IRA funding. GAO also analyzed and summarized selected agency review data and found these data to be sufficiently reliable for the purpose of describing the status of reviews with some limitations. For example, EPA and NTIA did not provide all requested data. GAO also interviewed selected agency officials for each of the objectives.
Here’s what they found…
The Infrastructure Investment and Jobs Act (IIJA) and Inflation Reduction Act (IRA) provided budget authorities for transportation, infrastructure, and energy projects. Together, the four agencies selected for this review—the Environmental Protection Agency (EPA), Department of the Interior (Interior), National Telecommunications and Information Administration (NTIA), and Department of Transportation (DOT)—obligated a majority of their IIJA or IRA funding.
- IIJA. Of the approximately $574.7 billion provided to them, the four agencies obligated about 76 percent and disbursed about 54 percent of obligated funds for fiscal years 2022 to 2025.
- IRA. Of the around $53.7 billion provided to them, EPA, Interior, and DOT obligated about 72 percent and disbursed about 60 percent of obligated funds for fiscal years 2022 through 2025. In July 2025, Congress rescinded $6.4 billion of the three agencies’ unobligated IRA funds per agencies’ data.
Most selected agencies reported developing new processes to review IIJA and IRA funding in response to dozens of executive orders issued starting January 20, 2025. The agencies used a variety of approaches to implement reviews, such as searching for terms from executive orders like “diversity” or “environmental justice.” Senior leadership for all selected agencies made final decisions on whether awards should be approved without modification, modified (funded with changes to terms or amounts), or canceled (discontinued), according to selected agencies’ officials and documentation. The extent to which agencies continued to obligate and disburse IIJA and IRA funds during their reviews varied within and across agencies.
As a result of their reviews, selected agencies reported approving about 9,500 awards ($128 billion) and canceling about 800 ($17.8 billion). More than 2,500 awards were pending a decision ($33.6 billion) as of varying dates shown below.
Status of Selected Agencies’ Infrastructure Investment and Jobs Act and Inflation Reduction Act Funding Reviews, as of the Dates Indicated
I found this fascinating, from the Minnesota Supercomputing Institute,,,
What if artificial intelligence could help scientists discover new drugs for healthy aging?
Thomas Hodder, the Life Sciences & Engineering Runner-Up at this year’s Research Computing Exhibition, is developing new ways to identify compounds that may promote healthy aging. Hodder created an AI-powered imaging process that measures heat stress resistance, a trait strongly linked to longevity, in the microscopic worm C. elegans
Traditional lifespan experiments can take weeks and require significant manual effort. Hodder’s approach automates much of the process, allowing researchers to rapidly screen thousands of compounds and identify promising candidates more efficiently. Using this method, his team discovered 14 previously unknown drugs that extend worm lifespan.
One of the biggest challenges was developing and validating the assay itself. Hodder spent nearly two years refining the method before it was ready to detect potential drug candidates. Throughout that process, he relied on high-performance computing resources from the Minnesota Supercomputing Institute to process large imaging datasets, test software, and store research data.
