What is the difference between download and upload speed?

Download speed is data coming to your device — streaming video, loading web pages, receiving files. Upload speed is data leaving your device — video calls, cloud backup, file sharing, livestreaming. Most home broadband plans are asymmetric: faster download than upload. Check both when evaluating a plan.

Is 100 Mbps fast enough for working from home?

Yes for most users. A video call (Zoom, Teams) requires 3–5 Mbps upload and download per stream. 100 Mbps download is comfortable for 2–3 people working and streaming simultaneously. Upload speed matters more for remote work — check your plan's upload spec, as some cable plans offer only 10–20 Mbps upload on a 100 Mbps download plan.

What is latency and why does it matter?

Latency (ping) is the round-trip time for a small data packet, measured in milliseconds. Low latency (<20 ms) is critical for gaming and real-time video calls. High bandwidth with high latency (e.g., 100 Mbps but 80 ms ping) still feels sluggish for interactive use because each request-response cycle takes longer. Satellite internet often has 600+ ms latency even with adequate bandwidth. For gaming and video calls, prioritize low latency alongside sufficient bandwidth.

Internet Speeds Explained: Mbps, MB/s, and What Your Plan Actually Delivers

One of the most common points of confusion about internet speeds is the Mbps vs MB/s distinction — a factor of 8 that makes your connection look faster than it is. ISPs advertise in megabits per second because the numbers are 8× larger; download managers report in megabytes per second because that maps to file sizes. This guide explains the difference, how to calculate real download times, and what speeds actually mean for everyday use cases.

Published March 20, 2026

Key takeaways

Mbps = megabits per second. MB/s = megabytes per second. 1 MB/s = 8 Mbps.
ISPs advertise in Mbps because the numbers are 8× larger. Download managers show MB/s.
A 100 Mbps plan delivers ~12.5 MB/s of file download speed.
For 4K streaming: ~25 Mbps per stream. HD video: ~5–8 Mbps. Video calls: 3–5 Mbps up.
'Up to' speeds are theoretical maximums. Real-world speeds depend on network congestion, Wi-Fi overhead, and server capacity.

Bits vs bytes: the root of the confusion

There are 8 bits in 1 byte — this single fact explains nearly all the confusion around internet speed units. Networking standards measure bandwidth in bits per second (or kilobits, megabits, gigabits). File sizes, however, are measured in bytes (or kilobytes, megabytes, gigabytes). The formula is simple:

MB/s = Mbps ÷ 8

A 100 Mbps connection transfers 100 ÷ 8 = 12.5 MB/s. A 500 Mbps connection transfers 62.5 MB/s. A 1 Gbps (1,000 Mbps) connection transfers 125 MB/s.

ISPs advertise in Mbps because the numbers are 8× larger — '100 Mbps' sounds faster than '12.5 MB/s' even though they're identical. This isn't deceptive; bits-per-second is the technically correct unit for network bandwidth. But it creates confusion when users compare it to file sizes.

How to calculate download time

Formula

Time (seconds) = File size (MB) × 8 ÷ Speed (Mbps)

Examples:
• 1 GB file (1,000 MB) on 100 Mbps: 1,000 × 8 ÷ 100 = 80 seconds ≈ 1.3 minutes
• 4K movie (50 GB = 50,000 MB) on 500 Mbps: 50,000 × 8 ÷ 500 = 800 seconds ≈ 13.3 minutes
• 4K movie (50 GB) on 100 Mbps: 50,000 × 8 ÷ 100 = 4,000 seconds ≈ 67 minutes
• 10 MB email attachment on 10 Mbps: 10 × 8 ÷ 10 = 8 seconds

Note: these are theoretical maximums. Real downloads are typically 70–90% of advertised speeds due to protocol overhead, server limits, and network congestion.

What speeds do you actually need?

Speed requirements depend on what you're doing and how many people are using the connection simultaneously.

25 Mbps (the US FCC minimum broadband definition): handles HD streaming on 1–2 devices and light browsing. Tight for a household with multiple devices.

100 Mbps: comfortable for a family of 4 — HD streaming on 2–3 devices simultaneously, video calls, casual gaming, and general browsing. The sweet spot for most households.

200–500 Mbps: multiple 4K streams, working from home with video conferencing on multiple devices, online gaming, moderate file uploads.

500 Mbps–1 Gbps: large households with heavy simultaneous use, content creators uploading large video files, remote workers handling big data transfers, home NAS servers.

Upload speed matters too, especially for video creators (uploading to YouTube), remote workers (video conferencing, screen sharing), cloud backup, and gaming. Most home plans are asymmetric — faster download than upload. Symmetric fiber connections (same upload and download) are worth seeking out for intensive upload use cases.

Kbps, Mbps, Gbps — the full scale

Internet speeds use decimal (SI) prefixes, not binary:

1 Kbps = 1,000 bits per second (kilobits per second) 1 Mbps = 1,000,000 bits per second (megabits per second) 1 Gbps = 1,000,000,000 bits per second (gigabits per second)

For context:

Dial-up modem: 56 Kbps (0.056 Mbps)
Early ADSL: 1–8 Mbps
Modern cable/DSL: 50–500 Mbps
Fiber (FTTH): 100 Mbps–10 Gbps
Mobile 4G LTE: 10–100 Mbps (typical)
Mobile 5G: 100 Mbps–1 Gbps (typical)

The ISP 'gigabit' plan (1 Gbps = 1,000 Mbps) delivers a theoretical 125 MB/s — enough to download a 50 GB 4K movie in about 7 minutes under ideal conditions.

Why your speed test differs from your plan

Note

Several factors reduce real-world speeds below the advertised 'up to' figure:

Wi-Fi overhead: Wi-Fi adds protocol overhead and is subject to interference, distance attenuation, and channel congestion. A 1 Gbps fiber connection through a Wi-Fi 5 router may test at only 300–500 Mbps. Ethernet (wired) always delivers closer to the theoretical maximum.

ISP congestion: Shared infrastructure means speeds drop during peak hours (evenings, weekends). A provider advertising 500 Mbps may deliver 200 Mbps during prime time.

TCP overhead: The TCP/IP protocol adds approximately 5–10% overhead, meaning a 100 Mbps link achieves ~90–95 Mbps of usable throughput.

Server-side limits: The server you're downloading from may be throttled or overloaded. Speed test servers are optimized for throughput; real content servers may be slower.

For reliable testing: run speed tests via Ethernet, at different times of day, using multiple test servers.