Charter, Comcast, Rogers, others show off step-change performance improvements

Cable is in a pitched battle for broadband subscribers in the U.S.. It has been losing. On one side of cable, fixed wireless access (FWA) from mobile providers, with its “good enough” value proposition, has captured the bulk of broadband customers growth the past three years. On cable’s other flank, fiber-to-the-home (FTTH) outperforms cable, according to Ookla’s Speedtest® Connectivity Report for the United States in H1 2025, and fiber providers are holding their ground on subscribers. 

What’s cable to do? Fight back. Cable operators are deploying newer cable technology and reallocating frequency bands to differentiate their performance from FWA and close the gap with fiber, particularly in uplink speed and latency. The jargon of these efforts includes terms like DOCSIS 3.1, DOCSIS 4.0, mid-splits, and high-splits.

Key takeaways

• The upgrades are real and consumers can tell. Speedtest Intelligence® data confirms that major cable operators are actively upgrading their networks. The fingerprint of these upgrades is a clear, often multi-step, increase in median upload speeds, which is visible in market-by-market data from 2024 to 2025.
• Divergent strategies in the U.S. Operators are taking different paths. Comcast is pursuing a broad, nationwide rising tide mid-split upgrade to prepare its entire footprint for DOCSIS 4.0, with upload speed improvement everywhere. Charter is executing targeted, high-impact “surgical strikes” with high-split upgrades in specific markets yielding dramatic speed leaps. 
• In Canada Rogers showcases two systems. The legacy Shaw network in Western Canada was already upgraded, while Rogers is now actively bringing its Eastern Canada footprint up to that same high standard, creating a harmonized, next-generation network.
• The path forward. These mid-split and high-split upgrades are not the final destination but the crucial foundation. They re-engineer the network’s capacity, paving the way for the symmetrical, multi-gigabit speeds promised by DOCSIS 4.0.

Methodology

Using Ookla’s Speedtest® Intelligence data, we analyzed the performance of U.S. and Canadian cable providers, looking for evidence of these deployments. In particular, we  focused on changes in median uplink speed, year-over-year, from Q2 2024 to Q2 2025 across each provider’s larger markets. This analysis revealed a shift in upstream capacity that is a result of  of migrating from  a legacy “sub-split” architecture to the more advanced “mid-split” or “high-split” configurations that are part of the DOCSIS 3.1 cable standard and prepare the network for the DOCSIS 4.0 standard.

What are DOCSIS and Splits?

Data Over Cable Service Interface Specification (DOCSIS) is a telecommunications standard for high-bandwidth data transfer across cable television systems. The technology was initially optimized for downlink traffic to deliver television programming. With the rise of cable broadband, uplink capabilities were introduced to support interactivity.

Splits refers to how much spectrum is allocated to upstream traffic (splitting the spectrum between uplink and downlink). The sub-split upstream path is 37 MHz wide; mid-split upstream path is more than double that at 80 MHz wide; and high-split upstream path is nearly 200 MHz wide. There are technology and investment trade-offs between a mid-split and high-split deployment strategy, but fundamentally a wider path provides faster upload speeds, foreshadowing the results of this analysis.

Today, competitive pressure from fiber for faster, symmetrical speeds and lower latency is pushing cable technology forward. At the same time, consumer demand for high-performance gaming, seamless video conferencing, and content creator streaming is pulling the technology to evolve.

United States

Comcast (Xfinity)

The Speedtest user data provides a clear picture of a nationwide network upgrade and a successful rollout of mid-split architecture across Comcast’s footprint. Unlike the market-by-market approach seen with other providers, our data for Comcast suggests a sweeping, uniform upgrade program that has boosted median upload performance for customers across the country. This is a foundational step that directly supports Comcast’s DOCSIS 4.0 strategy and is a key element of its Project Genesis network initiative.

The nationwide mid-split upgrade

In Q2 2024, nearly every city examined showed a median upload speed in the ~23-24 Mbps range, the signature of a network operating on a narrow frequency band of a legacy sub-split architecture.

By Q2 2025, the picture had completely changed. Almost every market showed a jump in median upload speed to the ~40-42 Mbps range, an increase of roughly 75-80%. This is the fingerprint of a mid-split upgrade, which expands the upstream spectrum. This upgrade not only provides an immediate boost to upload speeds for DOCSIS 3.1 customers, but also reconfigures the network for DOCSIS 4.0 technology.

Here is a sample of representative markets demonstrating this consistent upgrade:

A first glimpse of Comcast DOCSIS 4.0?

While the mid-split upgrade is impressive in its breadth, the data for one city stands out:

• Colorado Springs, CO: 36.42 Mbps -> 63.86 Mbps

Colorado Springs’ Q2 2024 starting point was already higher than others, suggesting it was an early recipient of the mid-split. The subsequent jump to nearly 64 Mbps by Q2 2025 deviates from the ~40 Mbps norm seen elsewhere and could be early evidence of Comcast’s DOCSIS 4.0 deployment. In late 2023, Comcast announced Colorado Springs as one of its first three launch markets for “X-Class” symmetrical multi-gigabit speeds, along with select areas of Atlanta and Philadelphia. The higher median upload speed in Colorado Springs indicates that a growing number of customers adopting DOCSIS 4.0, lifting the city-wide median beyond what a mid-split upgrade alone can provide.

The other X-Class initial launch markets, Atlanta (41.40 Mbps) and Philadelphia (40.09 Mbps), still fall in line with the standard mid-split results for now, which is expected given the limited “select areas” footprint of the DOCSIS 4.0 rollout in these markets, per Comcast’s press release.

Comcast is preparing its entire network to compete head-on with fiber and FWA by lifting all boats with a mid-split and then by launching a new class of multi-gig symmetrical service.

Charter (Spectrum)

Speedtest Intelligence data from Q2 2024 to Q2 2025 reveals a targeted and significant deployment of high-split network upgrades in specific markets, aligning with Charter’s publicly stated network evolution plans.

Upgraded markets: the high-split transformation

In several key metropolitan areas, the jump in median upload speed between 2024 and 2025 was a transformative leap, often by a factor of 5x to 9x. This is the definitive signature of a high-split upgrade, which reallocates spectrum to create a much larger pathway for upstream traffic.

The data for the Dallas/Fort Worth metroplex and the Kentucky markets align directly with Charter’s announcements listing them as completed high-split markets. 

Reno, one of the earliest markets to be upgraded, already showed an upload speed over 100 Mbps in Q2 2024. In fact, examining the Speedtest data just prior to our Q2 2024 benchmark, we see that the upgrade came at the same time resulting in a 6x increase from February to June.

Markets awaiting upgrade

The majority of cities in the dataset, including Los Angeles (21.70 Mbps) and New York (21.06 Mbps), showed modest changes, suggesting they still operate on a legacy sub-split architecture. During Charter’s latest earnings call, President and CEO Chris Winfrey stated that “Step 1” of their upgrade plan was complete in approximately 15% of their footprint. This 15% figure suggests that the markets identified in our data with improved speeds are these “Step 1” locations. With 85% of Charter’s footprint remaining and upload speeds potentially ~7x faster, that’s a lot of upside.

Cox

The Speedtest Intelligence data for Cox Communications indicate a slightly more complex story than Charter’s, but with Cox not yet a public company (Charter merger is targeted for mid-2026), there isn’t a lot of publicly available information on the company’s deployment plans.. There appears to be a multi-stage upgrade strategy underway across their national footprint. Some markets are receiving initial mid-split upgrades, others are seeing those mid-splits mature, and a select few are now being pushed even further into high-split territory in preparation for DOCSIS 4.0.

Unlike the binary “upgraded or not” picture we saw with Charter, the changes in median upload speed for Cox reveals three distinct phases of network enhancement.

Phase 1: Initial mid-split deployment (sub-split to mid-split)

In some markets, there is a jump from the legacy sub-split baseline of ~10-15 Mbps, often doubling the median upload speed. This can represent the first step in modernizing the network by expanding the upstream spectrum.

• Gainesville, FL: 16.47 Mbps -> 34.97 Mbps (+112%)
• New Orleans, LA: 11.51 Mbps -> 23.76 Mbps (+106%)

These markets may be in the process of being upgraded, providing customers with a noticeable, though not yet final, improvement in upstream performance.

Phase 2: Mid-split maturation

The Speedtest Intelligence data for Cox indicates a multi-stage upgrade strategy with three phases of network enhancement.

• Phase 1: Initial mid-split deployment. Some markets show a jump from a legacy sub-split baseline, often doubling the median upload speed. For example, Gainesville, FL, increased from 16.47 Mbps to 34.97 Mbps (+112%).
• Phase 2: Mid-split maturation. Many cities, particularly in Arizona, California, and Nevada, were likely already operating with mid-split architecture in Q2 2024. By Q2 2025, these speeds saw a general uplift, with Phoenix, AZ, moving from 53.71 Mbps to 58.11 Mbps.
• Phase 3: High-split jump. In several markets, there was a large jump from an already-upgraded mid-split baseline to speeds approaching or exceeding 100 Mbps. This is the clear signature of a high-split upgrade, a prerequisite for Cox’s DOCSIS 4.0 path.

Cox’s phased approach allows the company to manage its capital expenditures while improving its customer experience along its DOCSIS roadmap.

Altice (Optimum)

Instead of pursuing a comprehensive DOCSIS 4.0 upgrade, Altice is executing a dual-pronged strategy: maximizing its existing DOCSIS 3.1 network while simultaneously building out fiber-to-the-home (FTTH).

Speedtest Intelligence data for Optimum markets shows a network in a steady state. From Texas to New Jersey, median upload speeds remained consistently in the 25-35 Mbps range, with no significant jumps that would indicate mid-split or high-split upgrades. For example, New York, NY, moved from 25.14 Mbps to 29.13 Mbps.

In fact, in its 2Q 2025 earnings report, Altice stated, “Mid-split upgrades on DOCSIS 3.1 network continue, and are expected to enable multi-gig speeds to a portion of HFC passings in 2026.” The company’s goal is to offer multi-gigabit speeds across 65% of its total service area by the end of 2028, using this combination of upgraded cable and new fiber.

Altice is making a calculated decision to invest capital in building new, future-proof fiber in targeted areas, while making more modest, incremental upgrades to their existing cable plant.

Mediacom Communications (Xtream)

Regional provider Mediacom’s network appeared to be operating on a traditional low-split architecture as of Q2 2025. In Q2 2024, median upload speeds were consistently in the ~30-47 Mbps range. By Q2 2025, these speeds saw only minor changes, with Des Moines moving from 39.3 Mbps to 47.7 Mbps. There was no tell-tale jump to 60+ Mbps that would indicate a mid-split deployment.

This data suggests that while Mediacom has announced DOCSIS 4.0 trials, the foundational mid-split or high-split upgrades had not yet been deployed at scale. 

However, in May 2025, Mediacom talked about deploying high-splits in Des Moines. While too early to move the city-wide median in Speedtest data, a number of faster upload speed samples are beginning to appear in the data, signaling that a broader rollout may be imminent.

Canada

Rogers

The Rogers Speedtest data reveals a tale of two networks: the legacy network in Eastern Canada that is being upgraded to the new standard, and the mature, already long-upgraded network in Western Canada (the former Shaw network).

Eastern Canada: the upgrade program

Cities in Rogers’ legacy Eastern Canada footprint show the clear signs of a network in the midst of an upgrade cycle, comparing data for Q2 2024 with Q2 2025:

• Toronto, ON: 40.01 Mbps – 48.91 Mbps
• Ottawa, ON: 37.54 Mbps – 51.67 Mbps
• Mississauga, ON: 44.31 Mbps – 58.06 Mbps
• Brampton, ON: 47.68 Mbps – 69.69 Mbps
• Markham, ON: 45.91 Mbps – 75.96 Mbps

These Ontario markets in Q2 2024 demonstrated median upload speeds in the 35-50 Mbps range. While this is better than a legacy sub-split, it may indicate a mid-split deployment that is still in its early stages. The substantial and consistent growth across all these cities by Q2 2025 is the evidence of Rogers bringing its Eastern network up to the capabilities of the West.

Western Canada: the Shaw legacy

In cities that were part of the Shaw network prior to the acquisition, the Q2 2024 data shows a network that was already highly advanced.

• Calgary, AB: 96.16 Mbps – 137.29 Mbps
• Winnipeg, MB: 101.07 Mbps – 138.06 Mbps
• Edmonton, AB: 96.71 Mbps – 125.68 Mbps
• Vancouver, BC: 87.69 Mbps – 105.91 Mbps

Speedtest data for Q2 2024 for cities like Calgary, Edmonton, and Vancouver already reflected a post-upgrade network, indicating a very mature mid-split network. Shaw was proactive with its network enhancements well before the Rogers acquisition was finalized. According to Shaw’s Principal Network Engineer in October 2021, the company began its mid-split upgrade program back in 2017. By late 2021, more than 90% of that work was already finished, with the entire project slated for completion by early 2022. 

Looking back to this time period in Calgary, Shaw more than tripled the median upload speed in one year:

• Q2 2021: 19.93 Mbps
• Q4 2021: 47.97 Mbps
• Q2 2022: 70.99 Mbps

Calgary stands out as a top performer, which makes sense as it was the former headquarters for Shaw as well as the location for Rogers’ successful DOCSIS 4.0 trial that achieved 1 Gbps upload speeds.

The Rogers data provides an interesting textbook case of a post-merger network integration and upgrade strategy. That is, bringing its Eastern Canada plant up to the mid-split standard already established by Shaw in Western Canada. And, a future strategy directly informed by its technology partnership with Comcast. Rogers is establishing a nationwide mid-split foundation, which is the necessary prerequisite for DOCSIS 4.0 and symmetrical multi-gigabit speeds.

Cogeco

Operating in Ontario and Quebec, Cogeco is employing a market-by-market mid-split upgrade strategy. In Q2 2024, every city examined showed a median upload speed capped at approximately 30-34 Mbps, indicative of a low-split architecture. By Q2 2025, several cities showed a significant jump while others remained unchanged. Upgraded markets include:

• Oakville: 34.5 Mbps – 70.6 Mbps (+104%)
• Burlington: 34.2 Mbps – 59.9 Mbps (+75%)

This data provides an example of a phased network evolution. The cities with ~60-70 Mbps upload speeds are where Cogeco has performed the mid-split upgrade, creating the upstream capacity needed to eventually offer the performance of DOCSIS 4.0.

Videotron

Videotron’s performance in Québec and Ottawa reveals a different story. In Q2 2024, median upload speeds were between 34 Mbps and 47 Mbps. By Q2 2025, those speeds saw only a minor increase, with medians sitting between 42 Mbps and 49 Mbps.

The data suggests Videotron operated a high-performing DOCSIS 3.1 network on a traditional low-split architecture as of Q2 2025. This makes Videotron a possible “before” snapshot—a baseline of what a highly optimized low-split DOCSIS 3.1 network looks like just prior to initiating the mid-split or high-split evolution.

Summing uplink

The evidence from Speedtest Intelligence is that the cable industry in the U.S. and Canada is not standing still. Faced with fierce competition, operators are making significant commitments to evolve their cable networks. However, they are not all following the same blueprint.

The data reveals a strategic divergence. On one path, operators like Comcast and Charter are going all-in on DOCSIS, betting that multi-billion-dollar upgrades can extend the life of their networks for another decade and allow them to compete head-on with fiber. Yet even they differ in tactics, with Comcast pursuing a broad, foundational upgrade while Charter executes targeted, high-impact deployments.

On another path, operators like Altice in the U.S. and another operator in Mexico (examined in this research, though not presented here) are hedging their bets. They are choosing to invest capital in building new, future-proof fiber networks while performing more modest, incremental upgrades to their existing cable plants. This two-tiered approach suggests a financial calculation that, in some areas, a full DOCSIS 4.0 upgrade is less attractive than a long-term fiber overbuild.

These upgrades represent the most significant architectural change to the cable network in over a decade. They are the essential groundwork for DOCSIS 4.0, which promises to finally deliver the symmetrical, multi-gigabit speeds needed to achieve performance parity with fiber. The question is not if cable will respond to its competitors, but how effectively and how quickly. Can DOCSIS technologies truly compete with the speed of light?