News Reporter |
News Editor | Wen Shuqi
Less than a year ago, Lenovo was still trimming its server business. Now, this segment—long a drag on profits—has become the company’s confidence booster in the AI boom.
On August 13, Lenovo Group released its Q1 FY2026/27 results. Quarterly revenue hit $26.943 billion, up 43% year-over-year—the fastest quarterly growth in nearly five years. After stripping out non-cash and non-operating items like fair value changes on warrant liabilities, adjusted net profit attributable to shareholders reached $1.075 billion, surging 176% YoY.
Under Hong Kong Financial Reporting Standards, Lenovo recorded a net loss of $609 million attributable to shareholders for the quarter, mainly due to a $1.69 billion fair value loss on derivative financial liabilities tied to warrants, not from operational changes.
This quarter’s momentum also pushed Lenovo to revisit its earlier medium-term targets. Chairman and CEO Yang Yuanqing said at the results call that the company had set a goal of $100 billion in annual revenue within two years—but at the current growth pace, “we’re hopeful we can hit it this year.” Lenovo had also pledged to lift net margin to 5% in three years and 7–8% in five years; Yang noted that timeline might accelerate too.
Across its three business groups, the Intelligent Devices Group (IDG)—covering PCs, phones, and tablets—posted revenue of $17.106 billion, up 27% YoY. The Solutions & Services Group (SSG) brought in $2.884 billion, up 28%, with an operating margin of 24.2%. But the biggest shift came from the Infrastructure Solutions Group (ISG), which for the first time looks less like Lenovo’s old server business.
Last fiscal year, Lenovo spent $285 million restructuring this division, trimming legacy compute organizations and costs while redirecting resources toward AI—and it only just returned to profitability the prior quarter. Back then, the question was whether the profits could hold. This quarter, ISG revenue hit $8.51 billion, nearly doubling YoY, with operating profit of $777 million versus a loss of about $86 million a year earlier. Operating margin swung from -2% to 9.1%.
Still, this can’t be chalked up purely to AI server fireworks. Yang told reporters at the results call that AI remains “a small slice” of ISG revenue, with traditional compute still dominating. But on the order pipeline front, AI server pipeline has ballooned from roughly $21 billion last quarter to $54 billion—a 157% sequential jump.
Lenovo isn’t alone in this surge. Dell, another PC maker, got there earlier. Dell’s Q1 FY2027 revenue hit $43.8 billion, with AI-optimized server revenue at $16.1 billion, up 757% YoY. Its Infrastructure Solutions Group brought in $29 billion with $3.1 billion in operating profit. Dell previously disclosed an AI server backlog of $51.3 billion.
These orders are bringing in a new breed of customer. A wave of Neoclouds—companies specializing in GPU compute—are popping up on server vendors’ client lists. They’re leaner and more agile, bulk-purchasing GPUs and data center resources, then renting out compute to model companies and enterprise clients.
Last fiscal year, Yang noted that large cloud providers alone couldn’t meet demand, and the emerging Neoclouds were growing even faster. Lenovo, with its global manufacturing footprint and customization know-how, moved quickly to seize the opportunity. This quarter, Lenovo said it delivered an “AI factory” with roughly 7,000 GPUs to a new-age cloud service provider.
These orders are also making servers look less like standardized machines of the past. Lenovo is leaning into its ODM+ model, Neptune liquid cooling tech, and “global resources, local delivery” manufacturing system. It has further split ISG into two lines: cloud infrastructure and enterprise infrastructure. Both grew 98% YoY this quarter—the former riding hyperscaler and Neocloud expansion, the latter increasingly targeting enterprise AI.
That second line aligns with a thesis Yang has hammered home over recent quarters: AI compute will shift from training to inference. Last quarter, he estimated 70–80% of AI infrastructure was still for training, with the long-term mix likely flipping to inference-heavy. This time, he described the current split as closer to “fifty-fifty,” and predicted it could ultimately settle at “80% inference, 20% training.”
The business logic is fairly straightforward: training giant models is concentrated among a handful of companies, but once models are actually deployed by businesses and individuals, inference happens across far more data centers, edge nodes, and endpoint devices.
Industry data at least partially backs the idea that inference is driving more traditional server demand. TrendForce’s server market research also notes that growing inference workloads are pushing cloud providers to expand general-purpose server deployments, not just keep buying GPU servers.
This is the backdrop for Lenovo’s ongoing “hybrid AI” narrative. Yang believes the AI frenzy of the past two years centered on public internet data and foundation models; the next big wave comes from private data that enterprises and individuals haven’t fully tapped. This quarter, SSG’s AI services revenue grew triple digits YoY, with managed services and projects & solutions accounting for 62.4% of the segment’s revenue.
But the same AI boom is creating headaches on Lenovo’s other flank.
AI servers don’t just gobble up GPUs—they devour HBM, DRAM, and NAND too. As memory makers shift more capacity to servers and high-value products, the RAM needed for PCs and phones gets pricier. TrendForce projects traditional DRAM contract prices will rise another 13–18% sequentially in Q3 this year, with NAND Flash up 10–15%. Its research points to AI server demand as a key pillar holding up memory prices.
Against this backdrop, Lenovo’s IDG still grew revenue 27% to $17.106 billion this quarter, with operating profit up 27% to $1.21 billion and operating margin steady at 7.1%. PC and smart device revenue rose about 30%, tablets grew 83%, and smartphones 15%. Global PC market share stood at 24.2%, up 0.5 percentage points YoY.
Yang attributes this resilience to supply chain muscle. Lenovo procures components for PCs and phones as well as infrastructure like servers, giving it greater scale. It also retains a fairly complete R&D, manufacturing, and supply chain ecosystem, allowing faster pass-through of back-end cost changes to front-line prices. The company has long-term supply agreements with some vendors, with memory sources spanning markets outside the U.S., including Korea and China. His read: as long as AI demand stays elevated, memory supply tightness will persist at least through this year and next.
AI PCs are also being bet on as the next growth driver. Lenovo’s global AI PC market share hit 25.1% in Q1. Liu Jun, president of Lenovo China, told reporters that AI PCs now account for more than half of Lenovo’s quarterly PC shipments in China, and the latest generation personal AI agent, Tianxi, is becoming a system running across PCs, phones, and tablets. Yang also confirmed plans to launch computers powered by NVIDIA RTX Spark later this year, enabling more complex AI agents to run locally.
But Lenovo isn’t the only one here. NVIDIA has already announced that the first wave of RTX Spark laptops and compact desktops will also come from Dell, HP, Microsoft Surface, ASUS, and MSI.
After the earnings release, Lenovo’s Hong Kong-listed shares spiked over 22% at one point. Bloomberg noted that the market has clearly re-evaluated Lenovo’s valuation logic this year—fewer investors see it as just a traditional PC maker, and more are pricing it on the growth prospects of an AI infrastructure company.
