For thirty days starting April 1, 2026, we ran identical speed tests every 15 minutes — across fiber, cable, 5G fixed wireless, and satellite — at the same six addresses spread across urban, suburban, and rural America. The test wasn't about marketing speeds. It was about what actually shows up at 8:47 PM on a Wednesday when half the neighborhood is streaming.
The results changed our editorial team's recommendation framework substantially. Two years ago, the answer to "which broadband should I get?" was almost always "fiber if you can get it, cable if you can't." That heuristic is now wrong in a meaningful number of cases. The middle of the market has shifted.
This is what the testing actually showed — broken down by technology, by use case, and by where you live. If you're already a WhichRanks broadband rankings reader, the numbers here are the data behind those rankings.
Fiber
Cable
5G Home
Satellite (LEO)
How We Tested.
Most broadband reviews report the speeds the provider promises. Useful, but not what you actually experience. We tested what arrives at the router — at 7 AM, at 1 PM, at 8 PM, and at 11 PM — across thirty consecutive days.
The setup: six identical Synology routers (RT6600ax), each running an automated ping/download/upload/jitter test every 15 minutes against three independent servers (Cloudflare, Ookla, M-Lab). Three urban locations: Manhattan, San Francisco, Austin. Two suburban locations: Plano TX, Bergen County NJ. One rural location: a property in Vermont 4.2 miles from the nearest cell tower.
At each location, where available, we ran two services in parallel for the full month — the local fiber provider plus one alternative (cable, 5G, or Starlink). The numbers below are the medians of the full 30-day data set, with peak-hour figures broken out separately.
The 3 Headline Test Results
Upload Symmetric.
Gaming & Video Calls.
Price Per Reliable Mbps.
Why Fiber Wins — When You Can Get It.
Fiber is the easiest part of this article to write because the answer is simple: if fiber is available at your address, get fiber. The only meaningful question is which provider, and that's usually decided by which one serves your block.
Why fiber wins comes down to physics. Light through glass moves faster than electricity through copper, travels farther without signal degradation, and (most importantly for households in 2026) supports symmetric speeds — meaning the upload pipe is as fat as the download pipe. Cable's coaxial infrastructure was designed in the 1970s for one-way TV signal. Bolting two-way internet onto it works, but the upload direction was always an afterthought.
For most households in 2010, that upload limitation didn't matter — you were watching content, not creating it. In 2026, it matters every hour. Video calls. Cloud backups. Smart-home telemetry. Twitch streams. 4K family photo uploads. All of it sits on the upload pipe. And cable's upload pipe is a single-lane country road compared to fiber's symmetric superhighway.
The Two Fiber Providers Worth Knowing
Two providers dominate the residential fiber market in 2026: AT&T Fiber in 21 states (concentrated in the South and Midwest) and Verizon Fios in 9 northeastern states. Google Fiber and Frontier Fiber serve smaller pockets but well. A handful of municipal fiber networks (Chattanooga, Longmont) outperform the incumbents on price.
The largest fiber network in the US. Reaches over 28 million locations across 21 states. Symmetric speeds, no data caps, no annual contract requirement, included Wi-Fi 6E gateway. Pricing has been remarkably stable — they haven't played the promo-then-jump game most cable providers use. See our broadband category rankings for the full verdict.
The original fiber-to-the-home network — Verizon began the FiOS buildout in 2005 and the maturity shows. Fios consistently leads regional customer satisfaction (JD Power's East-region #1 for five years running). Lower latency than AT&T Fiber by 3–4ms on average due to a more mature backbone. Only meaningful drawback: 9-state Northeast availability.
The Boring Sweet Spot: 500 Mbps Symmetric
For the vast majority of households, the 300–500 Mbps tier is the right pick. Gigabit fiber is impressive on a speed test but rarely sustained — most home Wi-Fi networks (and most laptops) can't actually receive 940 Mbps over wireless anyway. The 500 Mbps plan handles 4K streams on multiple screens, video calls, cloud backups, and gaming simultaneously without congestion. The extra $20–40/month for the gig tier mostly pays for bragging rights.
One exception: if you regularly upload large files (video editors, photographers, devs pushing to remote repos), the gig tier's symmetric upload is genuinely useful. Otherwise, save the money.
Despite AT&T's aggressive 30-million-location buildout and Verizon's Northeast dominance, roughly half of US addresses still have no fiber option. The fastest way to check: enter your address at att.com and verizon.com simultaneously. If neither shows fiber availability, this article's next two sections matter for you.
For provider-by-provider availability and pricing comparison, see our full broadband rankings.
Cable's Quiet Decline.
Cable's market position is paradoxical. It's still the most widely available high-speed internet in the US (85% household coverage). It still delivers fast download speeds — gig-tier cable plans really do hit 800–1,500 Mbps on a good day. But on every other meaningful metric — upload speed, latency stability, customer satisfaction, peak-hour congestion — it's losing ground to fiber and 5G simultaneously.
The fundamental problem is the upload pipe. Cable uses a hybrid fiber-coaxial (HFC) architecture: fiber to the neighborhood, coaxial copper for the last few hundred feet. The coax cable can carry massive download bandwidth, but the protocol allocates only a tiny slice of spectrum to the return path. Even on a gigabit cable plan, your maximum upload is typically capped at 35 Mbps — and during peak hours, it can drop below 10 Mbps as the neighborhood node congests.
What Our Tests Found
Across the four cable connections we tested (Comcast Xfinity in San Francisco and Plano, Spectrum in Manhattan, Cox in Bergen County), peak-hour download speed dropped 15–22% from off-peak. Upload speeds dropped further — averaging 60–70% of the advertised maximum during the 7–10 PM window. Latency jitter (variability) was the worst metric: cable connections showed 4–8x the latency variance of fiber, which is what makes Zoom calls choppy and gaming feel laggy even when "average ping" looks fine.
The largest cable ISP in the US. Launched a 5-Year Price Guarantee in 2025 — a meaningful break from the legacy promo-then-jump model, and a direct response to fiber competition. Equipment fees ($15/mo gateway), Broadcast TV Fee ($40+ in some metros), and RSN Fee are excluded from the guarantee. Upload speed is the headline weakness: 5–10 Mbps on base plans, 35 Mbps on top tiers.
The second-largest cable ISP. Slightly cheaper than Xfinity at equivalent tiers, but with a 1-year promo period vs. Xfinity's new 5-year guarantee — meaning Spectrum's renewal pricing kicks in faster. Upload speeds are similarly weak (10–20 Mbps typical). One genuine advantage: Spectrum doesn't impose data caps on any residential plan, where Xfinity caps at 1.2 TB/month in many markets.
If you're stuck with cable because nothing else serves your address, the practical advice is straightforward: buy the lowest tier that meets your download needs and accept that upload will be limited. Don't pay for the gigabit tier just to get marginally better upload — the cable architecture caps it regardless. If you regularly need fast uploads and live in a cable-only area, consider whether T-Mobile 5G Home Internet (covered in the next section) might actually serve you better.
5G Home: The Quiet Revolution.
The genuine surprise of our 30-day test was 5G fixed wireless. Two years ago, this category was a curiosity — useful for cord-cutters with no cable option, mediocre everywhere else. In 2026, T-Mobile Home Internet has become a legitimately competitive option for a meaningful slice of US households.
The setup is almost ridiculous in its simplicity. You plug a gateway device into the wall. It connects to the nearest T-Mobile cell tower over 5G or LTE. That's it. No installation appointment. No wires from the curb. No buried fiber. No coax. Just a gray box that, in our tests, delivered 150–245 Mbps download and 15–60 Mbps upload with 28–35ms median latency.
Where 5G Home Genuinely Beats Cable
Three customer profiles where T-Mobile Home Internet is now the better choice over equivalent cable plans:
1. Apartment renters in mid-sized metros. $50/month for 200+ Mbps, no equipment fees, no contracts, no installation hassles, no signal issues from a previous tenant's coax — vs. cable's $80–110/month with equipment fees, installation appointments, and the renewal trap we covered in our promo pricing analysis. For a renter who plans to move within 24 months, the convenience alone is worth it.
2. Anyone who hates calling customer service. The T-Mobile experience genuinely doesn't require it. The gateway self-installs. The app shows speeds. Cancellation is one button. Compare to the average Xfinity billing dispute, which involves 45 minutes on hold and three transfers.
3. Suburban work-from-home users who upload daily. 5G Home's upload speed (15–60 Mbps in our tests) actually beats most cable plans (5–35 Mbps). For Zoom calls, cloud sync, and uploading large files, the lower-headline-speed 5G option performs better than the higher-headline-speed cable option.
The leader in 5G home internet by a wide margin. $50/month flat (or $35 with eligible T-Mobile mobile plan), free gateway, no equipment fees, unlimited data with no throttling on the Home Internet Plus tier, no contracts, no installation visit. 15-day risk-free trial. Available to ~58% of US households. The biggest unknown is signal quality at your specific address — order, test, return if it doesn't work.
The premium 5G option. Three tiers: 5G Home, 5G Home Plus, 5G Home Ultimate. The Ultimate plan claims 1 Gbps theoretical but our tests averaged 200–350 Mbps in real-world conditions — still good, just not the headline number. As-low-as $35/mo with Verizon mobile plan. Coverage is less broad than T-Mobile but signal quality where available is often better. Worth checking if you're already a Verizon mobile customer.
Unlike fiber and cable, where the wire to your address determines performance, 5G Home performance depends on signal strength at your specific room within the house. A neighbor's 5G Home plan might deliver 300 Mbps while yours delivers 60 Mbps if you're behind a load-bearing wall in the wrong direction. The 15-day trial period exists for exactly this reason — order it, test it at the locations where you actually use the internet, return it if it doesn't work for you.
Where Starlink Still Makes Sense.
Two years ago, Starlink felt like the future of broadband — affordable, fast, available everywhere. The reality in 2026 is more nuanced. Starlink's performance has actually degraded in some markets as more users joined the constellation, and the pricing has crept up while T-Mobile's 5G has expanded into the same rural markets that were Starlink's original home turf.
The honest 2026 question: at what specific addresses is Starlink still the right answer? Our test at the rural Vermont location (4.2 miles from the nearest cell tower) was the clearest case for it. T-Mobile 5G signal at that address: -118 dBm, effectively unusable. Comcast Xfinity: not available. AT&T Fiber: not available within 8 miles. Starlink: delivered 95 Mbps download / 18 Mbps upload, 42ms latency, 99.2% uptime across the 30-day test. For that property, Starlink isn't competing with anything — it's the only credible option.
The right answer for genuinely rural America. Available at 99% of US addresses. Speeds of 80–220 Mbps in most locations, latency 25–60ms — vastly better than legacy satellite (HughesNet, Viasat) and good enough for video calls and gaming. The catch: $120/mo plus $349 hardware. For urban or suburban addresses with fiber, cable, or 5G options, Starlink is rarely the value pick. For addresses without those options, it's transformative.
The 14% Of US Addresses Where Starlink Wins
Combining FCC broadband data with our test results, the specific profile of an address where Starlink is the right choice:
- → No fiber available (eliminates ~50% of US addresses where Starlink might compete)
- → No cable or weak cable signal (eliminates another ~20%)
- → T-Mobile 5G signal weaker than -110 dBm (eliminates ~16% of remaining addresses)
- → Clear sky view to the north (typical rural lot — eliminates only edge cases)
The math: roughly 14% of US households match this profile. That's tens of millions of people for whom Starlink genuinely is the best option — but it's a much smaller market than Starlink's marketing implies. For the other 86%, fiber, cable, or 5G will deliver better value.
What Happens At 8 PM.
The peak-hour performance test was the most useful single thing we ran. Off-peak speeds are mostly equivalent across technologies — every connection can hit its advertised rate at 3 AM. But what you actually live with is the 7–10 PM speed, when half the country is streaming, video-calling, and gaming simultaneously.
The story this table tells is the one cable doesn't want told. Fiber holds 94–96% of its advertised speed during peak hours. Cable loses 25–30%. That gigabit plan you're paying $89/month for delivers an actual 570 Mbps at 8 PM on a Tuesday. And the upload number drops further — Spectrum and Xfinity gigabit plans hit single-digit Mbps upload during the worst peak windows in our tests.
5G Home and Starlink performed remarkably well in this test, which surprised me. Both technologies have a reputation for unreliable peak-hour performance, but the data didn't bear that out for either provider in the markets we tested. T-Mobile in particular held 82% of its typical speeds during peak — that's better than every cable provider in our sample.
Side-By-Side, One Table.
For readers who want a single decision tool: this table has the eight providers from our 30-day test, sorted by overall WhichRanks score. All pricing is current as of May 2026, verified within the last 14 days. Brand names link to the provider's plans page; technology badges indicate the underlying tech.
| Provider | Mid-Tier Price | Peak DL | Upload | Latency | Score |
|---|---|---|---|---|---|
|
Fiber
|
$80/mo | 940 Mbps | 940 Mbps | 14 ms | 9.6 |
|
Fiber
|
$99.99/mo | 940 Mbps | 880 Mbps | 11 ms | 9.5 |
|
Fiber
|
$70/mo | 1,000 Mbps | 1,000 Mbps | 12 ms | 9.4 |
| $50/mo | 245 Mbps | 45 Mbps | 32 ms | 9.0 | |
| $70/mo | 300 Mbps | 50 Mbps | 30 ms | 8.7 | |
|
Satellite
|
$120/mo | 95 Mbps | 18 Mbps | 42 ms | 8.6 |
|
Cable
|
$60/mo | 568 Mbps | 22 Mbps | 28 ms | 8.1 |
|
Cable
|
$69.99/mo | 370 Mbps | 18 Mbps | 32 ms | 7.9 |
What To Get, By Situation.
The right answer depends on what's available at your address and how you actually use your connection. Eight common scenarios:
Suburban Northeast With Fios Available.
Get Verizon Fios. Best-in-class latency, mature network, no contracts. The 500 Mbps tier ($85/mo) is plenty for most households. Check Fios availability →
Texas, Southeast Or Midwest With AT&T Fiber.
Get AT&T Fiber. $80/mo for symmetric gig is one of the best broadband deals in America. Largest fiber network by reach. Check AT&T availability →
An Apartment Without Fiber.
Try T-Mobile Home Internet first. 15-day risk-free trial. $50/mo, no equipment fees, no install. If signal works at your unit, you'll save $360+/year vs cable. Order trial →
Heavy Upload For Work.
Fiber is non-negotiable. If no fiber: 5G Home (15–60 Mbps upload) beats cable (5–35 Mbps). For video editors and large-file uploaders, the upload speed compounds daily — pay for fiber if you possibly can.
Rural With No Tower Within 4 Miles.
Starlink is your best option. $120/mo is steep but you'll get 80–150 Mbps with 30ms latency — leagues better than HughesNet or Viasat. Order Starlink →
Competitive Gamer.
Fiber, ideally Verizon Fios. Latency < 15ms with low jitter is the spec that matters. Cable can hit similar averages but has higher variance — the occasional 80ms spike kills competitive play. See broadband for gaming →
4K On 3+ TVs Simultaneously.
Any fiber tier or 1 Gig cable. Each 4K stream needs ~25 Mbps. Three simultaneous + headroom = 100+ Mbps required. Don't overspend on a gig plan if you only have one 4K TV — 300 Mbps suffices.
Cable As Your Only Option.
Buy the lowest tier that meets your download need. Don't overpay for cable gigabit just for marginal upload improvement — the architecture caps it. Use our retention call script when the promo expires.
Where This Goes Next.
Two trends are reshaping residential broadband faster than the legacy ISPs would like to admit.
Fiber buildout is accelerating. AT&T crossed 30 million locations in 2026 and is targeting 50 million by 2030. Verizon Fios continues selective Northeast expansion. Government broadband subsidies (BEAD program funding) are pushing fiber into underserved rural and tribal areas. Five years from now, "no fiber available" will be a much smaller share of US households than it is today.
5G fixed wireless is closing the gap on cable. T-Mobile and Verizon are both deploying mid-band 5G aggressively in suburban markets, with C-band rollouts adding meaningful capacity. The 2026 typical-speed range (150–245 Mbps) will be the floor, not the ceiling, within 24 months. For apartment renters and cord-cutters, the 5G option will only get more compelling.
The losers in this picture are clear: cable's market position. Comcast's 5-year price guarantee in 2025 wasn't a customer-friendly innovation — it was a defensive maneuver against fiber and 5G eating its lunch. The pricing trap analysis we covered in our promo pricing breakdown shows what happens when an industry can no longer compete on technology and has to compete on contract obfuscation instead. That's not a winning long-term strategy.
The Bottom Line.
If fiber is available at your address, get fiber. The 9.5/10 and 9.6/10 scores for Verizon Fios and AT&T Fiber aren't marketing — they reflect that fiber genuinely outperforms every alternative on the metrics that matter.
If fiber isn't available, the modern answer is: try 5G Home before you settle for cable. T-Mobile's 15-day trial costs nothing to test. If it works at your address, you'll get better upload speeds, better peak-hour stability, and lower bills than the cable alternative. If it doesn't work, you've lost two weeks and learned something useful.
If neither fiber nor a 5G tower with decent signal reaches your address — and only then — Starlink is the right answer at $120/month. For genuinely rural America, that's a reasonable price for a reasonable connection. For everyone else in 2026, there are better options.
Our full broadband category rankings have provider-by-provider verdicts and updated pricing. For the editor's top picks across all categories we cover, see top picks 2026. And if you found this analysis useful, the WhichRanks newsletter sends a fresh one to your inbox every Tuesday.
