The answer

Only some of them. HPE is solving the problems of the upper midmarket, firms with 1,000 to 4,999 employees, because those problems look like enterprise problems: virtualization exits, on-premises AI and large network refreshes. For SMBs with 1 to 99 employees and for the core midmarket with 100 to 999, HPE is offering enterprise answers to questions those firms are not asking, priced and packaged in ways that work against how they buy.

The evidence comes from two places: what Techaisle's SMB and Midmarket Datacenter Solutions Adoption Trends Study (N=2,857) shows SMBs need, and what HPE's product line, settlement obligations and partner terms show it is building.

The smallest SMBs skipped the generation HPE is selling them

Techaisle's data describes firms with 1 to 99 employees as Skip-Generation SMBs. These firms carry almost no legacy infrastructure and little or no VMware exposure, and they are not modernizing on-premises estates because most never built one. They are moving from minimal on-premises infrastructure straight to cloud and SaaS, bypassing the server-and-hypervisor stage the midmarket is now unwinding.

HPE's answer for these firms remains a server catalog. The tower and entry rack servers are reliable, and they serve the business that needs a file server and a domain controller in a back office. That business is a shrinking share of the segment. The sovereignty pressure that pulls larger firms back on-premises barely registers here: data sovereignty and IP control is an architectural driver for 22% of SMBs with 1 to 99 employees, against 82% of the upper midmarket.

For decades, midmarket infrastructure procurement followed the hardware-aging cycle: IT departments replaced servers and switches as they depreciated. In the upper midmarket, that cycle has broken. According to the Techaisle SMB and Midmarket Datacenter Solutions Adoption Trends Research Report, 57% of the overall SMB market has a major infrastructure refresh underway or scheduled within 12 months, and in the upper midmarket the primary catalyst is now proactive workload enablement rather than hardware aging.

The data gravity of artificial intelligence is driving a hardware supercycle. 32% of upper midmarket firms are refreshing to meet new workload requirements, including AI, against 20% replacing hardware at end of life. In the core midmarket, end-of-life replacement still leads at 28%. The supercycle is creating distinct architectural tracks based on organizational maturity, and major ecosystem players such as Dell, Lenovo, Cisco, AWS, and Google Cloud need to reevaluate how they position their portfolios for each. The disjointed accumulation of isolated products has created an integration gauntlet, and midmarket customers are demanding cohesive, full-stack orchestration.

For most of the past two years, a midmarket firm that decided to run AI on its own infrastructure faced an awkward wait. It could approve the budget, pick the use case, and win over the CFO, then wait months, sometimes close to a year, for the GPU servers to arrive. By the time the hardware landed, the business case had aged, and the sponsor had often moved on.

Lenovo AI Express goes after that wait directly. Customers choose one of three validated Lenovo Hybrid AI Factory configurations built with NVIDIA, and Lenovo commits to order-to-ship in 15, 20, or 25 business days. Lenovo built the program for customers of every size, from SMB to enterprise. I focus on the midmarket here because that is where a delivery date changes the most.

For a midmarket buyer, that date matters more than anything on the data sheet. An AI project in a firm of this size usually rests on one sponsor and one budget cycle, and a delivery date lets that sponsor show results inside the fiscal year that funded the purchase. That is often what separates a second phase from a canceled one. Plenty of vendors have built AI reference designs, so the contest has been about who has the best architecture. Lenovo is the first I have seen put a delivery date at the front of the sales conversation, which moves the contest to who can actually deliver.

Key Takeaways

• The advantage is not the AI PC itself. It is the breadth of device categories Lenovo can integrate and manage through a common software and services layer, which is what lets a single agent work across an entire fleet. A vendor focused primarily on PCs has fewer categories to work across.
• The FIFA World Cup 2026 proved the model at scale. More than 26,000 Lenovo and Motorola devices across three countries, deployed and managed as one fleet, with FIFA highlighting Lenovo’s rapid deployment and managed lifecycle approach as helping accelerate operational readiness in weeks rather than months.
• The constraint is go-to-market, not portfolio. MSPs influence 61% of SMB PC decisions, yet only 34% of SMBs say their MSP explains the business value of an AI PC. Closing that gap is the highest-return move available to Lenovo today.

For the past two years, almost every conversation I have had about the AI PC has been a conversation about a single device. How many TOPS. Which NPU. How the battery holds up running a model locally. Vendors brief on it, partners repeat it, and buyers listen politely before asking the only question they actually care about: what any of this does for their business on Monday morning.

I have come to think the industry has been measuring the wrong thing, and that the mistake is more basic than any argument about silicon. We keep asking a single device to deliver value that is not created on a single device. Almost nobody does their job on one screen anymore.

Watch how a piece of work moves through a company today. A proposal gets drafted on a laptop, discussed in a message thread on a phone, approved on a tablet in the back of a taxi, and then picked apart the following week on a workstation by someone in finance who was never in the original meeting. No single device holds that story. The work lives across all of them, and so does the context that explains it.

That distinction matters enormously once you put an AI agent into the picture, because an agent is only as useful as the context it can actually reach. An agent that lives on the laptop and nowhere else is reasoning about a fraction of what happened. It will summarize the document but miss the decision, because the decision was made on the phone. Techaisle research consistently shows buyers reporting the same frustration in different words, and it is the reason so many AI PC deployments have been underwhelming in practice even when the hardware was perfectly capable.

"We keep asking a single device to deliver value that is not created on a single device," Anurag Agrawal .