Techaisle Analyst Insights

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 .

First, the verdict: the new XPS 14 is a laptop I recommend, and highly. I have a complicated history with the XPS name, though, and so does Dell.

I have been using the XPS brand laptops since 2012, when the first XPS 13 arrived as one of Intel's Ultrabooks, a $999 Windows answer to the MacBook Air. I was not always an easy sell on where Dell took the line next. When 2-in-1s were the fashion, I stood at an Intel offsite and argued for two hours, with Techaisle data behind me, that the category would struggle. I had tried the contraptions, including Dell's original XPS Ferris-Wheel, and none of them convinced me to give up a clamshell. Then a few years later Dell sent me an XPS 13 2-in-1; I put it through a road test, and it made a convert of me. That is the thing about this line. It has a habit of winning arguments I walked in ready to lose.

By 2020, I was calling the XPS 13 perfection personified. I opened the 9300, set it on my desk, and let it sit for three days before I could bring myself to use it. For years, this was the one Windows laptop I could hand to someone without a caveat.

Then Dell decided the name had to go. In 2025, XPS became Dell Premium, the top rung of a new consumer ladder of Dell, Dell Plus, and Dell Premium, while the Pro and Pro Max labels went off to the commercial and workstation machines. It all looked tidy on an org chart and meant nothing to the person actually shopping for a laptop. I sat through more than one briefing where I watched a room full of smart people explain a naming system that answered a question nobody had asked. The machines stayed good throughout.

They just lost the one word that told you so at a glance.

This year Dell admitted it made a mistake and brought XPS back. New design, new silicon, same three letters. I have spent the last few months with the new XPS 14, and I have enjoyed nearly all of it.

Worldwide IT spending by firms with 1 to 4,999 employees will reach US$1.667 trillion in 2026, excluding communication services, and the majority of it will go to IT services rather than to technology products. A market of that size, spread across every economy and every industry, sets the direction for commercial IT rather than following it. These firms are now spending more on the implementation, integration, management, and security of technology than on the technology itself, and the margin between the two is wide and widening.

That composition is the product of two forces working against each other. AI is pulling money up and forward, into software, infrastructure, and services that were not in the budget a year ago. Cost is pulling the other way, as component inflation, tighter budgets, and a higher cost of capital are pushing firms to defer what they can and to rent what they cannot. That second force is the quieter one, and it explains the tilt toward services better than any capability argument does. Buying an outcome instead of an asset moves cost from the balance sheet to the income statement, and it moves operational risk from the firm to the provider. In a year of expensive capital and unforgiving threats, that trade is worth paying for, which is why the money is moving toward services even where the technology itself is cheap.

Within services, the mix has shifted. Maintenance, support, and break-fix, the labor of keeping systems alive, once defined the SMB services market. The money is now concentrating in consulting, integration, and putting AI into production. Transformation work has overtaken recurring management, and it is not close.

Michael Dell opened Dell Technologies World with a line that sounded like theater but was actually a strategy: just as electricity transformed the world when it left the power plant, AI will transform the world when it leaves the screen. With intelligence becoming infrastructure, the job now is to make it real, local, secure, and useful, whether that is on an oil rig, in an ambulance, or on the factory floor.

The most revealing moment came a day later, when Jeff Clarke admitted that his own engineers burned through a month's worth of allocated tokens in a few hours. This happened not because something broke, but because it worked perfectly. Put those two moments together, and you have the entire event's thesis. Michael Dell named the destination (intelligence everywhere it is needed), while Clarke named the bill that arrives when you get there. Ultimately, what Dell announced was not a refresh cycle; it was a bet on where intelligence physically lives, and who pays the meter to run it.

The number that should reset every infrastructure budget

From the keynote stage, Jeff Clarke cited figures that framed everything that followed: token prices have fallen roughly 80% year over year, yet consumption for reasoning has surged 320-fold. Furthermore, inference, not training, now accounts for nearly two-thirds of all AI compute. Whatever the underlying sources of this data, the direction is indisputable and directly mirrors what Techaisle has been tracking from the buyer side all year.

Reading those numbers together leads to an unavoidable conclusion: the unit cost of intelligence is collapsing, yet total spend is accelerating. This is the exact pattern Techaisle named Token Shock. We've seen this curve before with bandwidth, storage, and compute, where cheaper units unlock so much new consumption that the overall bill climbs anyway. What sets this era apart is the sheer speed, as no one has seen a cost curve bend this quickly.

The strategic consequence, and the line Clarke delivered that should be sitting in every CFO conversation, is that as agents take on more cognitive work, costs migrate from headcount to tokens. Historically, cognitive work scaled with human hours; if you wanted more analysis, you hired more analysts. Agentic AI has broken that ratio entirely. Techaisle data puts a number on how far it has already shifted: the Agent-to-Human Ratio has reached 144-to-1 in the midmarket and 59-to-1 in small businesses. With the agentic workforce already deployed at that density, it's alarming that most of the operating models meant to govern it still assume a payroll rather than a token budget.

Dell's actual announcement was an answer to "Where"

Across both keynotes, one question sat underneath every announcement: where should a given token run?