Federal prosecutors in California have charged a man with shipping roughly $300 million worth of advanced Nvidia GPUs to China without the required export approvals — allegedly routed through intermediaries to support what the reporting dryly calls 'super intelligence' development. Whatever happens in court, the operational fact underneath the headlines is uncomfortable: hundreds of millions of dollars of hardware moved through a supply chain, and nobody reconciling that supply chain caught it in time.
Now bring it home. Nobody on your team is moving pallets of datacenter GPUs out a loading dock. But the same failure mode exists in nearly every mid-size IT environment: hardware enters, moves through, and leaves your organization without IT — the department accountable for it — holding a reliable record. A department head buys five laptops on a corporate card. A contractor's workstation never gets an agent. A server gets racked in a closet in 2022 and appears in nobody's inventory. Then legal, an auditor, or a cyber insurer asks a simple question: what hardware do you have, where is it, and who put it there? And the answer takes three weeks of spreadsheet archaeology.
This article is about closing that gap with RMM-driven asset discovery and inventory — and why monitoring and remote management living in the same platform is the only practical way to keep an accurate picture of what you actually manage.
The Problem in Depth: Why IT Loses Track of Its Own Hardware
Your RMM only knows what has an agent. Most platforms — ConnectWise, Ninja, Datto, take your pick — build their inventory from agent check-ins. If procurement happens outside IT, there's no agent, so there's no device. It gets no patches, no monitoring, no backup, no inventory record. From your console's perspective, it doesn't exist — until a user opens a ticket about a laptop your helpdesk has never seen.
Spreadsheets are not a CMDB. Every shop has the inventory.xlsx on a file share, last meaningfully updated before the last two rounds of org changes. It's stale the day after someone touches it, and nobody trusts it — least of all the person who has to defend it in an audit.
Procurement and operations live in different worlds. The Nvidia case is alleged to involve hardware moving through channels that bypassed the people and systems responsible for oversight. Translate that pattern down to your organization: purchasing happens in finance, deployment happens in IT, and the two never reconcile. Finance has an invoice for a workstation your RMM has never seen. Or worse — your RMM sees a device that matches no purchase order at all, which raises an even less comfortable question.
The audit moment. Concrete scenarios any sysadmin or MSP tech will recognize:
- Your cyber insurance renewal asks for a complete hardware and software inventory. Your RMM reports 340 agents. The spreadsheet says 290 devices. Which number do you certify in writing?
- An auditor asks you to enumerate every system with a specific hardware component — discrete GPUs, certain NICs, specific firmware — for licensing or compliance reasons. How long until you can answer?
- MSP scenario: a client asks where a batch of decommissioned workstations went. The previous provider's documentation says 'disposed — see vendor.' Which vendor?
- Finance flags $180K of hardware purchases in Q3. You walk the floor and find 12 of those machines with no agent, no monitoring, no backup, and no patching.
What it costs:
- Unmanaged endpoints are, statistically, unpatched endpoints. They're the machines that get flattened by the ransomware wave your patched fleet shrugged off — and then bridge it onto your network.
- Audit and insurance responses stretch from hours into weeks, and an inaccurate certified answer creates legal exposure, not just embarrassment.
- Technician hours burn on manual reconciliation — RDP machine to machine, run msinfo32, type serial numbers into a spreadsheet at 9pm.
- MSPs eat margin: discovery during onboarding was scoped as a fixed-price project, and undocumented client gear is the classic scope-creep killer.
The gap exists because the tooling is siloed. The RMM tracks agents, the helpdesk tracks tickets, procurement tracks invoices, and nobody owns reconciliation between them. Legacy RMM architecture treats inventory as a byproduct of agent check-ins rather than a first-class dataset that gets continuously reconciled against reality.
How AlertMonitor Closes the Gap
Discovery that doesn't depend on agents. AlertMonitor's network discovery scans your subnets and identifies every device — workstations, servers, printers, switches, firewalls — including anything without a managed agent. The delta between 'devices on the network' and 'devices in inventory' is your blind spot, surfaced as a concrete list instead of a suspicion.
Full hardware inventory on every managed endpoint. Manufacturer, model, serial number, CPU, RAM, disk, GPU and driver versions, installed software, OS build — collected automatically at every check-in and stored in the same platform as your alerts and your tickets.
New-device alerts. When a device appears on a monitored network, AlertMonitor raises an alert the same day, with the device details attached. You decide: enroll it or investigate it. Either way, it never lives unpatched for eight months because nobody noticed.
Asset-tied ticket history. Because the helpdesk is in the same platform, every ticket ever opened against a machine is attached to that asset. When an auditor or a client asks for the history on a specific box, it's one click — not a search across three tools and someone's memory.
Remote sessions and scripted audits. When you need to verify something on a machine — confirm the chassis serial, check a GPU model, pull driver versions — you open a remote session or push a script to a device group from the same console. Results feed back into the monitoring timeline, so the record shows who ran what, where, and when.
Scheduled compliance reports. Inventory by site, by client, by hardware class — generated on a schedule and delivered automatically. Nobody builds it by hand the week before the audit.
The old workflow: monitoring console in one tab, RMM in a second, helpdesk in a third, spreadsheet in a fourth, technician reconciling between them by eye. In AlertMonitor, the device, its inventory record, its alerts, and its ticket history are one object on one timeline. Teams moving off spreadsheet-and-agent-only workflows typically cut audit prep from weeks to hours — and the 'wait, we own that machine?' moments stop happening entirely.
Practical Steps: Get an Accurate Hardware Picture This Week
1. Run a discovery sweep before you touch the spreadsheet
In AlertMonitor: Network Discovery → run against every subnet → compare discovered devices against the agent inventory. Investigate everything that appears in one list but not the other. Do this first — the spreadsheet is the least trustworthy source you have.
2. Pull hardware inventory with PowerShell
For Windows fleets, this collects the fields auditors and asset registers actually ask for:
$servers = Get-Content '.\targets.txt'
$report = foreach ($c in $servers) {
$cs = Get-CimInstance -ClassName Win32_ComputerSystem -ComputerName $c
$bios = Get-CimInstance -ClassName Win32_BIOS -ComputerName $c
$cpu = Get-CimInstance -ClassName Win32_Processor -ComputerName $c
$gpu = Get-CimInstance -ClassName Win32_VideoController -ComputerName $c
[PSCustomObject]@{
Hostname = $cs.Name
Manufacturer = $cs.Manufacturer
Model = $cs.Model
SerialNumber = $bios.SerialNumber
CPU = $cpu.Name
RAM_GB = [math]::Round($cs.TotalPhysicalMemory / 1GB, 1)
GPU = ($gpu.Name -join '; ')
}
}
$report | Export-Csv '.\hardware-inventory.csv' -NoTypeInformation
In AlertMonitor, push this as a script to a device group instead of running it machine by machine — results land in the platform timeline and attach to each asset record automatically.
3. Audit GPUs and driver versions specifically
If you have GPU workloads — VDI, ML boxes, rendering workstations — know exactly what's deployed and which driver each one runs:
Invoke-Command -ComputerName (Get-Content '.\gpu-hosts.txt') -ScriptBlock {
if (Get-Command nvidia-smi -ErrorAction SilentlyContinue) {
nvidia-smi --query-gpu=name,driver_version,memory.total --format=csv,noheader
} else {
'No NVIDIA tooling detected'
}
} | Export-Csv '.\gpu-driver-audit.csv' -NoTypeInformation
On Linux hosts:
sudo dmidecode -s system-manufacturer
sudo dmidecode -s system-product-name
sudo dmidecode -s system-serial-number
lspci | grep -Ei 'vga|3d'
nvidia-smi --query-gpu=name,driver_version --format=csv,noheader 2>/dev/null || echo 'no nvidia-smi present'
4. Make new devices alerts, not surprises
Enable new-device alerts per site or per client. A device that appears on the network and isn't claimed by the agent inventory within a day should generate a ticket automatically. That workflow is what catches the contractor laptop before it becomes an unpatched resident of your network.
5. Tie procurement to enrollment
One process rule: no hardware deploys without an asset record. Invoice lands → device is enrolled → discovery confirms it's reporting. For MSPs, make discovery the first step of every onboarding — not the client's own device list, which is wrong more often than it's right.
6. Reconcile on a schedule, not on demand
Schedule the AlertMonitor inventory report monthly, and diff it against finance's fixed-asset list quarterly. Two hours of calm review per quarter beats three weeks of panic when an auditor is on the calendar.
The Bottom Line
The Nvidia export case is a story about hardware moving through a system nobody was reconciling. Your environment doesn't have to be. Agent-based inventory alone will always have a blind spot exactly where it matters most: the devices IT never touched. Discovery scanning, alerts on the delta, hardware inventory captured at every check-in, and a helpdesk tied to the assets — that's the difference between answering an auditor in an afternoon and explaining to your boss why the number you certified last year was wrong.
Related Resources
AlertMonitor RMM & Remote Management AlertMonitor Platform Overview Book a Demo RMM & Remote Management Resources
Is your security operations ready?
Get a free SOC assessment or see how AlertMonitor cuts through alert noise with automated triage.