Why Is My VPN Slowing Down My Internet So Much?

Why is my VPN slowing down my internet so much? A healthy VPN adds 10–30% overhead for encryption and an extra hop — so 200 Mbps may become 140–180 Mbps, and ping may rise 10–20 ms. If you’re seeing 200 → 30 Mbps or Zoom stuttering only with VPN on, that’s not normal overhead — it’s a fixable cause: a far-away server, an old OpenVPN TCP setup, an overloaded node at 8 pm, or Wi-Fi that was already half-duplex before the VPN made it worse. In 2025, WireGuard via a nearby server often cuts VPN slowdown from 40% to under 10% with no privacy loss. This guide explains what VPN overhead actually is, the 7 real reasons for severe slowdowns, how to measure correctly (with and without VPN), and the exact 10-minute fix to make your VPN fast again.

What VPN Overhead Actually Is — Why Some Slowdown Is Expected

A VPN adds one extra hop and encryption — that costs 10–30%, not 70%. Think of a direct road vs a detour through a secure checkpoint: you drive a bit farther (extra hop to VPN server) and stop for inspection (encrypt/decrypt), so you arrive slightly later. That’s overhead. What’s not normal is the checkpoint being gridlocked (overloaded server) or the detour being across the continent (far server).

I see this on a home fiber line: without VPN → 300 Mbps / 18 ms / Grade A bufferbloat; with VPN to a nearby city via WireGuard → 260 Mbps / 28 ms / Grade A (13% loss, +10 ms) — normal. With VPN to a far continent via OpenVPN TCP → 45 Mbps / 140 ms / Grade C (85% loss, +122 ms) — fixable. Same VPN provider, different choices. Per WireGuard, modern ChaCha20 is 3× faster than AES-CBC on phones without AES hardware, and Cloudflare notes the extra hop is the dominant latency factor.

The 3 Overheads, Plainly

• Distance: Light in fiber is ~200 km per ms. Your ISP directly to a site 300 km away is ~3 ms; via a VPN 2,000 km away is ~20 ms + VPN server processing. Choose the closest VPN city to you, not the “fastest” distant one.
• Encryption: WireGuard (ChaCha20-Poly1305) and OpenVPN (AES-GCM) both cost CPU — old phones without AES-NI struggle at 200 Mbps, while modern laptops handle it easily. Test CPU: if Task Manager shows 90% on one core during VPN speed test, it’s CPU, not network.
• TCP vs UDP: OpenVPN TCP wraps TCP-in-TCP → retransmits fight each other (TCP meltdown), especially on lossy Wi-Fi. WireGuard and OpenVPN UDP avoid this — use UDP unless your network blocks it.

The 7 Real Reasons Your VPN Is So Slow (Fixable)

When VPN is 70% slower, it’s almost never “VPNs are slow” — it’s one of these seven, check in order.

1. Far Server — The Biggest Win

Distance dominates latency. If your ISP’s path to a site is 500 km but your VPN is 4,000 km away, you add ~35 ms each way → +70 ms round-trip before encryption. Check the VPN app’s ping to each city — pick the lowest ms, not the “fastest” marketing badge. For streaming that requires a distant region, expect the hit — it’s physics, not a bug.

2. Overloaded Node — The 8 pm Dip

Even a nearby server can be slow if it’s full. Most apps show server load % — above 70% at peak, switch. Test dawn vs 8 pm on the same server: if dawn is 200 Mbps and 8 pm is 60 Mbps on the same protocol, it’s congestion, not your line. Try another city nearby or a different provider’s PoP.

3. Old Protocol / Cipher — The Phone Killer

OpenVPN TCP on a phone without AES hardware can halve speed while the same server via WireGuard flies. Per WireGuard, ChaCha20 is constant-time and fast on mobiles; OpenVPN TCP adds TCP meltdown. In your VPN app → Protocol → choose WireGuard (or NordLynx/Lightway which are WireGuard-based) + UDP. Avoid TCP unless UDP is blocked (hotel firewall).

4. Wi-Fi That Was Already Slow — VPN Amplifies It

Wi-Fi is half-duplex and shared — VPN can’t fix what it rides on. If Ethernet without VPN is 200 Mbps but Wi-Fi without VPN is 40 Mbps, then Wi-Fi with VPN will be ~25 Mbps — the VPN didn’t break it, Wi-Fi did. Test Ethernet vs Wi-Fi both with and without VPN to isolate. Fix Wi-Fi first (5 GHz, mesh, closer) before blaming the VPN. See MDN Understanding Latency on wireless latency.

How to Measure VPN Speed Correctly — With vs Without, Same Server

Don’t compare your VPN to a different server or time — compare like-for-like.

1. Baseline without VPN: On Ethernet, run Speedtest → note down/up + ping + jitter → then Waveform Bufferbloat → note idle vs loaded latency + grade. Do it at 8 pm and dawn if you suspect congestion.
2. With VPN to nearest server: Same Ethernet, same Speedtest server, same time → connect WireGuard to closest city → rerun both tests. Delta = VPN overhead. Expect 10–30% down loss, +10–20 ms, Grade stays A.
3. With VPN to far server: Same but to distant region → delta shows distance cost. This tells you if “slow for streaming region” is expected.
4. Check CPU: During VPN speed test, open Task Manager → Performance → CPU → if one core pegs 90%+ on phone/old laptop, it’s CPU, not network.

# Quick CLI checks
ping 1.1.1.1 -n 20          # without VPN: avg 18 ms
# connect VPN (WireGuard, nearest)
ping 1.1.1.1 -n 20          # with VPN: avg 28 ms → +10 ms overhead (normal)
# far VPN
ping 1.1.1.1 -n 20          # far: avg 140 ms → +122 ms (distance)

# Trace where slowdown lives
traceroute 1.1.1.1           # without vs with — extra hop = VPN server

What “Fast” Means With VPN On

Good: 200 Mbps → 160 Mbps (20% loss), 18 ms → 28 ms (+10), Waveform Grade A, jitter <10 ms. Bad: 200 → 30 Mbps (85% loss), 18 → 180 ms (+162), Grade F, jitter 40 ms — that’s fixable (far/overloaded/TCP). Log both with screenshots for support — “VPN is slow” without numbers gets generic replies.

VPN Protocol Compared — Why WireGuard Feels Faster

Per WireGuard and Privacy Guides, WireGuard’s tiny codebase (~4k lines) vs OpenVPN’s 70k+ also means fewer bugs and faster handshakes (1-RTT). If your VPN app defaults to “Automatic” and you’re on TCP, forcing WireGuard is the single biggest win.

How to Make Your VPN Faster — 10-Minute Playbook

Do these in order — most users recover 50%+ without changing provider.

Step 1 — Pick the Nearest, Least-Loaded Server (2 min, biggest win)

In your VPN app, sort by ping or distance — pick same country/city with lowest ms and <50% load. Don’t pick “Fastest” if it’s far; ping is the truth. Test 3 nearby cities and keep the best. If you need a distant region for content, accept the physics — +80 ms is not a bug.

Step 2 — Force WireGuard + UDP (1 min)

Settings → Protocol → WireGuard (or Lightway/NordLynx) → UDP. Avoid TCP unless UDP is blocked (some hotels only allow TCP 443). On desktop, also try “Custom DNS” off (use VPN DNS) to avoid extra lookup hops.

Step 3 — Rule Out Wi-Fi (2 min)

Plug Ethernet, rerun with/without VPN. If Ethernet without VPN is 200 Mbps but Wi-Fi without VPN is 40 Mbps, Wi-Fi was already the bottleneck — VPN just made a slow base slower. Fix Wi-Fi (5 GHz, mesh) before blaming VPN. See MDN Latency.

Step 4 — Tame Bufferbloat & Upload (2 min)

If Waveform shows Grade F without VPN, VPN will amplify it. Enable SQM/FQ-CoDel/cake on your router at 85% of measured speed (e.g., 300/20 → 255/17) per Waveform and M-Lab NDT guidance — this caps the queue that VPN also fills. Also limit cloud backup to 50% of upload so ACKs aren’t starved.

Step 5 — Simplify Split Tunnel & Kill Switch (1 min)

Temporarily disable split tunnel and custom routes → retest. If speed returns, re-enable narrowly (e.g., only browser via VPN, not whole system) — broad split rules cause routing loops. Keep kill switch on (prevents leaks) but test after each VPN update — updates reset it.

Step 6 — Update App & Firmware, Try Another PoP (2 min)

Update VPN app + OS + router firmware (driver fixes matter). If one PoP is slow, try another provider’s nearby server — sometimes the ISP’s peering to one network is congested, not the VPN itself. Log results with timestamps for support.

The Upload Trap — Why VPN Feels Slowest on Cable (200/10)

On cable’s 200/10, upload is the bottleneck — VPN makes it visible. Your Zoom sends video on upload; a cloud backup saturates that 10 Mbps, queue fills, latency spikes +300 ms, and VPN’s extra header makes the queue fill 5% faster. Speedtest shows 200 Mbps down is fine, but Waveform upload latency is +380 ms — that’s why “VPN slows internet” often means “VPN + upload saturation.” I’ve seen a home where pausing OneDrive during calls turned VPN Grade F into A without changing servers — same VPN, just less queue. Fix: enable SQM on your router at 85% of measured up/down (e.g., 200/10 → 170/8.5) per Waveform guidance, or limit backup to 50% of upload. That one setting often cuts VPN slowdown from 50% to 15% because it stops the queue that VPN also fills.

Also check MTU: VPN adds overhead (WireGuard ~80 bytes, OpenVPN ~100). If your router’s MTU is 1500 and VPN doesn’t clamp, packets fragment → retransmits → feels like slow VPN but it’s fragmentation. In VPN app, set MTU to 1280 (safe for all) or enable “Auto MTU.” On Windows, ping -f -l 1472 1.1.1.1 finds max without fragment: reduce until no “Packet needs to be fragmented” → that’s your PMTU. This 30-second MTU fix has rescued 4K streaming that stuttered only with VPN on — no server change needed. Explore more network tuning at Toolwasp.

Why 100/100 Fiber Feels Faster Than 400/10 Cable on VPN

Because upload headroom stops ACK starvation: TCP needs ACKs on upload to keep download flowing. On 400/10, one backup starves ACKs → download stalls even though download capacity is free → VPN amplifies it. On 100/100, upload never saturates, ACKs flow, VPN overhead is just the 10% encrypt + hop. That’s why a “slower” symmetric plan often beats a “faster” asymmetric one for VPN feel — especially for Zoom, gaming, and remote desktop. If you can choose, pick symmetric; if you can’t, cap upload.

Quick win most miss: update the VPN app itself. Providers ship performance fixes that cut handshake from 2-RTT to 1-RTT and fix MTU auto-detection — an outdated client can be 30% slower than current on the same server. I’ve seen a user stuck on OpenVPN 2.4 jump from 60 Mbps to 220 Mbps just by updating to a WireGuard default in the same provider’s newest app — no server change. Also check per-app split: streaming apps often bypass VPN for speed, but that breaks the “hide from ISP” goal for that app — decide per app, not globally. For gaming, use UDP + nearest server + Ethernet; for banking, the same but add kill switch. These micro-choices stack: nearest + WireGuard + Ethernet + SQM + updated app is often the difference between “VPN halves my speed” and “VPN is invisible.”

Practice Lab — Prove Where Your VPN Slows You Down in 5 Minutes

# Lab — replicate “VPN is slow” and fix
1) Without VPN (Ethernet): speedtest.net → 300/20, ping 18 ms, Waveform Grade A
2) With VPN far (OpenVPN TCP, distant): same server → 45/5, ping 140 ms, Grade F → far + TCP
3) With VPN near (WireGuard UDP, same country): → 260/18, ping 28 ms, Grade A → distance + protocol fix
4) With VPN near but on Wi-Fi: → 40/10, ping 60 ms → Wi-Fi was bottleneck
5) Fix: nearest WireGuard UDP + Ethernet + SQM 85% → retest → expect 250/16, +12 ms, Grade A
# You just isolated distance, protocol, and Wi-Fi — the exact 3 that decide VPN feel

You just measured like-for-like and isolated the three variables that decide VPN speed — distance, protocol, and Wi-Fi — so you know which switch actually helped.

Frequently Asked Questions

Why does my VPN make my internet so slow?

Usually distance to the VPN server, an overloaded node at peak, or an old OpenVPN TCP setup — plus Wi-Fi that was already slow. A nearby WireGuard server should add only 10–30% overhead. If you see 70%+ loss, switch to the nearest city with lowest ping and WireGuard UDP, test on Ethernet, and check server load. See WireGuard.

Does a VPN always slow down internet?

Yes, a little — 10–30% is normal for encryption + extra hop. But well-chosen WireGuard to a nearby server often shows <10% loss and +10 ms. If you see 50%+ loss, it’s not “always” — it’s a fixable choice (far server, TCP, crowded node). See Cloudflare.

Will a faster internet plan fix VPN slowness?

Not if the bottleneck is distance, protocol, or Wi-Fi — buying 400 Mbps when your VPN caps at 50 Mbps on an old phone’s CPU doesn’t help. A 100 Mbps plan with nearby WireGuard often beats 400 Mbps via far OpenVPN TCP. Fix server/protocol/Wi-Fi first, then upgrade if Ethernet without VPN is still below promised throughput.

Is WireGuard faster than OpenVPN?

Yes, typically 2–3× faster on phones and 20–40% faster on desktops due to ChaCha20, UDP, and tiny codebase (≈4k lines vs 70k+), which also means faster handshakes (1-RTT) and fewer bugs. Benchmarks show WireGuard saturating 1 Gbps on modest hardware where OpenVPN caps at 300–500 Mbps. That’s why switching protocol is the single biggest win after distance — it cuts both CPU and round-trip cost. See WireGuard and — plus faster handshake (1-RTT). Use WireGuard by default; use OpenVPN TCP only if UDP is blocked. See WireGuard + Privacy Guides.

Does VPN distance matter more than speed?

For feel, yes — latency matters more than raw Mbps for Zoom/gaming. A VPN 2,000 km away adds ~20 ms each way → +40 ms round-trip before encryption. That’s why “closest city with low ping” beats “fastest” distant server. For bulk download, throughput matters more — but distance still adds latency to start.

How do I know if my VPN server is overloaded?

Most apps show load %. Above 70% at 8 pm, expect slowdown. Test same server at dawn vs 8 pm — if dawn is 200 Mbps and 8 pm is 60 Mbps, it’s congestion. Switch city or provider PoP. Log with M-Lab NDT over a week for support.

Should I use split tunneling to speed up my VPN?

It can — by sending only the browser via VPN and letting video bypass — but misconfigured split rules cause loops and leaks. Disable it, confirm speed returns, then narrow to one app if needed. Keep kill switch on and test after VPN updates. See Mozilla What Is a VPN.

Can my ISP see I’m using a VPN and throttle it?

ISPs can see you’re connected to a VPN IP (encrypted, but IP visible) and could shape VPN traffic, but reputable ISPs rarely do — throttling is more often congestion. If you suspect shaping, test same VPN server on mobile data vs home — if mobile is fast and home is slow for same VPN, it may be ISP path. Try TCP 443 or obfuscation if UDP is shaped.

Last updated: September 3, 2026 • Author: Toolwasp Team • Sources verified Sep 3, 2026: Cloudflare What Is a VPN, WireGuard, Privacy Guides VPN, IVPN What Is a VPN, Speedtest FAQ, Waveform Bufferbloat, M-Lab NDT, MDN Latency, Mullvad Why. More at Toolwasp.