You start copying a 50GB video file to an external SSD through a USB-C hub.

File Explorer initially shows:

700 MB/s

Then:

500 MB/s

Then suddenly:

0 MB/s

A few seconds later, the speed may recover.

Or it may remain near zero, jump back to 300 MB/s, fall again, and continue in bursts until the transfer finishes.

It is tempting to conclude:

“The USB-C hub cannot sustain the transfer.”

Sometimes the USB path is the problem.

But a large-file copy that starts fast and then collapses can also reveal the SSD's write cache, thermal throttling, destination-drive behavior, shared USB bandwidth, or an actual USB reset.

The fastest way to diagnose it is not to focus on the highest number you saw.

It is to examine the shape of the entire transfer curve.

Read the Shape of the Transfer Curve

Different speed patterns point toward different parts of the storage path.

This distinction matters because:

A speed drop is not automatically a disconnect, and a disconnect is not automatically an SSD cache limit.

Curve #1: Fast Start → Lower Stable Plateau

Imagine:

First 15 seconds: 850 MB/s

then:

Rest of transfer: 280 MB/s

The copy did not fail.

The destination SSD may simply have moved from short-burst write performance to its lower sustained write performance.

Many SSDs use fast cache space to absorb incoming writes before committing that data to slower NAND. Micron's explanation of SSD dynamic write caching describes how DRAM or SLC cache can temporarily accept data at higher speed before data is committed to the underlying flash.

This means the first number File Explorer displays may represent:

burst performance

rather than:

the speed the SSD can sustain for 50GB or 100GB.

A similar principle can occur with consumer SSD technologies that use pseudo-SLC caching.

Once that fast cache is exhausted, the drive must write more directly to TLC or QLC NAND while also managing its internal flash.

The result can look dramatic:

800 MB/s → 300 MB/s

without any problem in the USB-C hub.

The clue

If the speed falls once and then remains relatively stable, investigate the destination SSD's sustained write performance before blaming the connection.

Curve #2: Fast → 0 MB/s → Fast Again

This pattern is different.

For example:

600 MB/s

→ 0 MB/s for 3 seconds

→ 350 MB/s

→ 0 MB/s

→ 400 MB/s

If the external drive remains mounted the entire time, the destination may be temporarily busy with:

cache flushing

flash management

garbage collection

or another storage operation.

Windows and the drive can both buffer writes, which means the File Explorer graph is not a direct real-time measurement of NAND writes every millisecond.

The important observation is:

Does the device remain connected?

If yes, do not immediately treat every zero as a USB disconnect.

Wait for the transfer to finish and compare the final average speed.

Also watch whether the external drive's activity LED remains active while File Explorer briefly reports zero.

Curve #3: 0 MB/s + Drive Disappears

Now the diagnosis changes significantly.

Suppose the transfer reaches:

0 MB/s

and then:

• the SSD disappears from File Explorer;
• Windows plays the USB disconnect sound;
• the drive reconnects;
• the copy fails;
• the drive letter disappears.

That is not normal SSD cache behavior.

The storage path itself has been interrupted.

Investigate:

SSD cable

→ Hub data port

→ Hub host connection

→ Power

→ Laptop USB-C port

If the same SSD is stable directly but resets only through the hub, the USB-C path becomes a much stronger suspect.

If the drive remains connected but Explorer itself freezes, use the separate File Explorer external-drive I/O timeout troubleshooting guide rather than treating every stalled copy as the same problem.

Curve #4: Speed Slowly Declines as the Drive Gets Hot

Another pattern looks like:

750 MB/s

→ 650 MB/s

→ 500 MB/s

→ 350 MB/s

over several minutes.

Unlike cache exhaustion, the slowdown may correlate more closely with temperature and sustained workload duration.

High-performance SSD controllers generate heat during long transfers.

Kingston's guidance on SSD thermal throttling and cooling explains that SSDs can reduce performance when temperature reaches a thermal limit, and that speed can recover after the drive cools.

To test for this pattern, let the SSD and enclosure cool completely.

Then repeat the same large-file transfer.

If you repeatedly see:

fast when cool → slower when hot → fast again after cooling

thermal behavior becomes much more plausible.

Avoid testing with the enclosure:

under the laptop

covered by cables

or:

stacked on another hot device.

Curve #5: Fast Alone, Slow When Another Drive Uses the Same Hub

This is a particularly important USB-C Hub pattern.

Imagine:

External SSD A → Hub

and:

External SSD B → Same Hub

Then you copy:

SSD A → SSD B

Both drives may individually advertise high speeds.

But they are still sharing the hub's upstream USB connection.

The data path effectively has to service both source and destination through the same upstream connection.

That is why:

SSD → Internal Laptop SSD

may remain fast,

while:

SSD A → SSD B on the same hub

drops sharply.

If the problem appears mainly when multiple high-speed devices are active, see the guide to why USB-C hubs slow down under a full peripheral load.

A hub adds ports.

It does not create unlimited independent bandwidth for every port.

Separate the Four Speed Ceilings

A large file copy is always limited by the slowest active part of the workflow.

Think of it as:

Source Read Speed

→ USB Link

→ Hub / Shared Path

→ Destination Sustained Write Speed

A transfer cannot run faster than the lowest ceiling in that chain.

Suppose:

The transfer may begin near:

800 MB/s

and later settle around:

280 MB/s

even though the USB connection never changed.

That is why the initial peak speed is often the least useful number in a long copy.

Record Four Numbers During One Large-File Test

Instead of watching File Explorer and guessing, copy one large file and record:

Then note one additional fact:

Did the drive disconnect? Yes / No

That single observation sharply separates:

storage performance behavior

from:

USB connection instability.

For cleaner results, use one large video or archive rather than thousands of small files.

Small-file workloads add directory, metadata, antivirus, and file-system overhead that make the curve harder to interpret.

Check the USB Link Before You Blame Sustained Storage Speed

There is another important scenario:

Your SSD is capable of 10Gbps.

But through the current connection it negotiates only:

480Mbps

or:

5Gbps.

In that case, the link itself is already limiting performance.

Use the actual USB link speed guide for Windows and macOS to distinguish:

Advertised Device Speed

from:

Negotiated USB Link

from:

Real File Throughput.

If the SSD is unexpectedly stuck at USB 2.0-class speed, follow the more specific external SSD stuck at 480Mbps troubleshooting workflow.

Do not mix that problem with an SSD that correctly negotiates 5Gbps or 10Gbps but loses speed only after its cache fills.

Run Three Copies to Find Which Side Is Slow

Use the laptop's internal SSD as a reference.

Copy 1 — External SSD → Internal SSD

This primarily tests the external drive's read side.

Copy 2 — Internal SSD → External SSD

This primarily tests the external drive's write side.

Copy 3 — External SSD A → External SSD B Through the Same Hub

This tests the shared hub workflow.

Suppose your results are:

That strongly suggests two separate effects:

destination sustained-write limit

and:

shared-hub workload

rather than one defective USB-C connection.

Free Space Can Change Sustained Write Behavior

If the SSD is almost full, repeat the test only after protecting your data and freeing meaningful space.

On many consumer SSDs, available free space can affect how much dynamic cache is available.

A nearly empty SSD and an almost-full SSD may therefore show different long-write curves.

Do not format the drive simply to improve a benchmark.

Instead, compare:

the same drive

same file

same connection

with enough free space available.

Also check the SSD manufacturer's documented sustained-write behavior rather than relying only on the large “up to” number printed on the product page.

When the USB-C Hardware Path Really Is the Bottleneck

Hardware becomes worth changing when your evidence looks like:

SSD direct → stable

but:

SSD through current hub → resets, negotiates lower speed, or stalls only under the same connection path.

For workflows using several USB-C storage devices, a USB-C hub with four USB 3.2 Gen 2 10Gbps data ports and up to 100W PD pass-through provides a storage-focused layout with clearly specified high-speed USB-C data connections.

The LENTION CE31s exposes four USB-C data ports rated up to 10Gbps and can provide additional power to attached peripherals when its PD input is connected.

That makes it relevant to workflows such as:

USB-C SSD + USB-C SSD + other high-speed USB-C storage

without repeatedly converting those devices to USB-A.

However:

Four 10Gbps ports do not mean four independent 10Gbps transfers can all run at full speed simultaneously.

They still share the hub's upstream connection.

If testing instead shows that the hub path is healthy but your current external M.2 enclosure becomes hot or limits sustained storage performance, a 10Gbps USB-C M.2 NVMe SSD enclosure with UASP and an aluminum heat-dissipation design can be used as a different diagnostic or storage configuration.

The current LENTION enclosure supports up to 10Gbps USB transfer and uses an aluminum shell with thermal material intended to transfer heat away from the installed SSD.

Again, neither product fixes:

SSD cache exhaustion

failing NAND

or:

file-system corruption.

The hardware recommendation only becomes relevant after testing identifies the USB path or enclosure as the limiting layer.

Use the Recovery Pattern to Identify the Bottleneck

A useful clue is what happens after you stop the workload.

The recovery behavior can be more diagnostic than the lowest MB/s number itself.

Quick Answers

Why does my large file copy start at 800 MB/s and later drop to 200 MB/s?

The destination SSD may initially be writing into a faster cache and then falling back to its lower sustained NAND write speed once that cache is exhausted.

Why does File Explorer sometimes show 0 MB/s and then recover?

The drive may temporarily be busy flushing cached data or performing internal storage management. If the device stays connected, a brief zero does not automatically mean a USB disconnect.

Is 0 MB/s caused by the USB-C hub?

Not necessarily. If the SSD disappears or reconnects at the same time, investigate the USB path. If it remains connected, investigate the destination drive, cache, thermal behavior, and workload.

Why does performance fall after several minutes?

If the slowdown correlates with rising temperature and performance returns after cooling, thermal throttling is a strong possibility.

Why is SSD-to-SSD copying slower when both drives are on the same hub?

Both storage devices share the hub's upstream USB connection, so the source and destination can compete for the same available path.

Does a 10Gbps hub guarantee 1,250 MB/s file transfers?

No. 10Gbps describes the signaling capability of the USB connection. Real file throughput also depends on protocol overhead, SSD performance, cache, enclosure, file system, and other active devices.

Final Takeaway

When a large file copy starts fast and then drops to 0 MB/s through a USB-C hub, do not judge the setup from the initial peak—or the lowest number alone.

Read the transfer curve.

Fast → Lower Plateau
usually points toward sustained SSD write performance.

Fast → 0 → Recover
can point toward cache flushing or drive-side work.

Gradual Decline
can suggest thermal throttling.

0 MB/s + Disconnect
pushes the investigation toward the USB path.

Fast Alone + Slow With Two Drives
points toward shared bandwidth.

The key rule is:

A large-file transfer is limited by the slowest active layer after the temporary burst performance is gone.

So diagnose:

Source

→ USB Link

→ Hub Path

→ Destination Cache

→ Sustained Write

→ Temperature

before replacing hardware.

Once you identify which part of the curve changes—and when—it becomes much easier to find the real bottleneck.