Interesting times require interesting measures. AI and data centers are taxing the hardware component ecosystem that supplies the device makers who deliver hardware for critical vertical market solutions. I love new hardware whether in a consumer or business context - a big reason why I’m here at Esper is because new hardware deployments are one of the best ways to roll out modern new solutions based on Esper’s device management infrastructure.

Should you just wait for prices to come down?

Given what a refresh costs right now, waiting it out looks tempting. Just be clear on what you're waiting for. Micron's CEO told investors in June that industry supply should "improve gradually in 2028" while also saying the company still has no line of sight on when supply catches demand. That's a business whose record margins depend on this shortage continuing, and even it won't call the end. And note the words: more supply, not lower prices.

When you should just refresh the fleet

Which means for dedicated device fleet operators, hardware refresh in 2026 and 2027 will be more expensive than previous refreshes. For some enterprises this is a budgetary annoyance. Modernizing the fleet is a bigger imperative with a clear ROI for the business that supplants budgetary concerns. The business case is clear. In that case go for it, work with your OEM, get the best deal you can, and refresh the fleet.

If you're running Windows 10 IoT LTSB 2016

But there is a particular cohort of fleets that is feeling a unique set of pressure - those that run Windows 10 IoT Enterprise LTSB 2016. These x86-based systems will no longer receive security updates from Microsoft after October 13, 2026. It is possible to continue to get security patch coverage which is only available directly from the OEM that supplied the hardware, at an escalating cost, and may not even be offered.

Yes, you have other Windows options such as moving to Windows 11 IoT (if your hardware even qualifies, and much of this era of hardware won't) or to Windows 10 IoT Enterprise LTSC 2021. But understand what you're signing up for: either path means reimaging every device in the fleet, with all the licensing costs and field workflow that entails. We've laid out the full comparison separately. In some cases the numbers crunch such that staying on Windows pencils out. Do it, but check out Esper’s device management for dedicated Windows fleets.

But what if you are in a constrained capital budgetary environment and your strategic roadmap demands your enterprise to move to an Android-based fleet before 2028?

Why we haven't pushed Firebolt before now

The proposition I give you is buy yourself a few years using a newly branded engagement Esper is calling Firebolt, which converts these Windows machines to Esper's Foundation for Android on x86. We haven't pushed Firebolt hard before now, because the math rarely called for it. When hardware is affordable and refresh cycles are healthy, refreshing is usually the right answer for our customers, and yes, for the OEMs and partners whose business is built around it. So we built the capability, we came through when an ISV needed endpoints on Android, and otherwise we let it sit. But the market has changed. Customers are facing hard choices, and we as the ecosystem have to get through it together especially when your AI brethren are securing 110% of the allocation and leaving little to nothing for the rest of us. Before we get into Firebolt, let's first introduce you to Esper Foundation and Esper Deploy.

What you'd be moving to

Esper builds and maintains the industry’s leading AOSP-based OS for dedicated device use cases - Esper Foundation for Android. We built it for various Arm-based platforms primarily serviced through a Generic System Image (GSI), but more important for this situation we also build and maintain for x86-based systems.

Our Foundation on x86 offers complete, operator controlled OTA infrastructure and builds quarterly OTA updates that incorporate Google’s latest security patches. In addition we track vulns in the Linux kernel sitting underneath Foundation and patch those as well. It requires Esper device management, but through this infrastructure IT Operators can also precisely control applying system updates using DevOps principles built into Esper.

Esper Deploy, our device management offering, is purpose built for dedicated device use cases. It includes a full REST API, CLI, and an on device SDK to super power the application layer. Furthermore, adding in Foundation delivers the Foundation SDK that enables deeper access to the Android framework than traditional AOSP, which can be used to design a cohesive experience for your users.

Foundation running in vertical markets is nothing new for Esper or the industry. Major enterprise brands run their front-line operations on Foundation today across restaurants and retail for POS, kiosk, KDS use cases in production, at fleet scale. We have support for many different peripherals and unique configurations required for these use cases - common examples include cash drawer and second display.

Does moving to Android mean slower devices?

One more thing that consistently surprises people: this is not a downgrade you tolerate to dodge an EOL date. Foundation is a lean, purpose-built AOSP OS carrying none of the accumulated weight of a general-purpose desktop OS and on 2016-era x86 hardware, the difference is immediate. Devices boot faster, applications feel snappier, and fleets that felt tired feel current again. I opened this post saying I love new hardware. This is the closest you'll get to that feeling without a purchase order.

Where Firebolt came from

This brings us to Firebolt. Many of the above Foundation on x86 deployments were brought to Esper using what we affectionately call the Windows to Android flip. This started in 2020 when I was running the Sales Engineering team so I was in the seat during both the sales process as well as the early rollout once we won our first deals.

Firebolt is two things: conversion tooling, and a tight presales engagement to quickly determine whether converting your specific fleet is feasible at scale.

I pose it that way as to deliver on the full implementation for the use case some NRE is typically required. This includes compatibility with all the permutations of hardware and firmware that compose your fleet, building any drivers required for the unique peripherals you need supported, and tuning the conversion tooling for your particular operational rollout should you require the tooling.

We have multiple options for you here. The sweetest is using your existing Windows RMM tool to push a purpose-built bootstrap file to your fleet and then remotely initiate the conversion at your pace. Depending on your requirements and existing installed base, you also choose whether Windows stays on the disk alongside Foundation or the device commits fully, and that gets vetted during scoping rather than at the bench. We can even enable you to toggle back and forth between Foundation and Windows to pilot in the field during off hours, noting this requires an existing Windows MDM in place.

Early on we found our customers wanted the safety net - after all in 2020 Foundation was new. These days Foundation has many industry miles on it so most customers skip it. The full breakdown of how a flip actually works, and what each path costs you, is here.

The beauty is Esper’s Seamless Provisioning is built into Foundation, so no matter what method you use to convert upon first boot into Foundation the device will be configured to spec using our Blueprint as the source of truth in our Esper Deploy device management. By the way, this is also true if the device is factory reset in the field any time after that - kind of an easy self-healing technique that requires no truck roll.

The big question is does Foundation support the core of what you need? For most fleets, yes. Where it gets interesting are the specialized peripherals, and there the key is the availability of a Linux driver - our engineers like to nerd out on these challenges. If a driver exists, support is straightforward. If not, it's a higher NRE because we have to scratch build it by reverse engineering, sometimes after chasing the vendor down with your help, sometimes with no help at all. Here is a more detailed look at the full accounting of what flips cleanly and what doesn't.

What scoping involves

This brings us to the NRE component. Yes, there are engineering costs. I always get asked how much? And the answer is always the same - it depends. We’ve seen our core Foundation on x86 work on the deployed hardware and peripherals straight up. But we have also had complex requirements that took quite a bit of engineering to get the solution working right. Where the NRE lands against the size of your fleet is what decides whether the economics work. That's a conversation, not a price list.

What do we require on your side in the presales process? A hardware inventory by model and revision. Why revision? We have seen situations where mid-stream component swaps require a different driver. Android is not Windows and driver support is build time not run time. So a display chip swap requiring a different driver during the life of a particular model can break Android. We have seen this happen. Android after all is about as purpose built as you can get! We need models and any SKU level revision information. For each of these variants what is very helpful to us is to generate SYSINFO exports you can share with us. This empirically tells us exactly the hardware you are running.

Then there are the peripherals. We need to know exactly the peripheral set you're running, cohorted by the common configurations in your fleet. Standard USB HID devices are easy. It's the older or more exotic gear that lands us back in the driver territory I described above, and that's where the NRE meter runs hottest, so flag those early.

From that we can start talking pricing which will encompass the NRE along with the ongoing subscriptions for Foundation maintenance and OTA along with Esper Deploy.

The best is if you are serious about this you are willing to loan us sample devices and peripherals for our lab where our engineers can perform initial validation. If that’s not possible, we can supply standard watermarked builds where you can run the initial smoke test in your lab. Frankly the best is when we can do both together and quickly run scoping and due diligence so we can get to the quote for you.

What goes wrong in the field

Are there potential hiccups here? Yes, what you see in the lab and pilot stores will not be the same as the field reality. BIOS drift, worn out SSDs initially blamed on a flaky Foundation build, a weird peripheral used by one franchisee you lost track of - we’ve seen a lot here and handled it all.

Where this doesn't fit

Are there limits to this? Of course. Foundation is primarily for fixed location devices - POS, Kiosk, KDS, Digital Signage. If your use case requires cellular modem, that is currently not supported. If the device is battery powered, Foundation does not offer power management at this level. But funny enough we have done battery powered devices and out-of-the box Foundation beat the other OS that we replaced in battery longevity. And the big one - are you dealing with an OEM/SecureBoot locked BIOS? If so you will need to find a way to unlock it at fleet scale.

If you're an OEM, ISV, or SI

Esper is a partner first company - we are a piece of the overall solutions stack spanning silicon vendors, ODMs, OEMs, ISVs, IHVs, VARs, SIs, and MSPs. Frankly these conversions work best when we all work together. If you are an OEM and have some customers in a tough spot with Windows 10 IoT Enterprise LTSB 2016 you’d like to help get them through so you are there for the next hardware refresh, we are happy to work together with you. If you are an ISV that needs endpoints to be on Android in order for it to work with your prospects, talk to us and let’s see what we can do.

In closing Microsoft’s EOL for Windows 10 IoT Enterprise LTSB 2016 is happening at a bad time relative to hardware refresh. As an enterprise customer you have options, and if you did not know about Esper Foundation with Firebolt we encourage you to consider it. Please reach out!