Featured Notes
When Can AI Software Stocks Move Off the Bottom? From Value Reset to Market Repricing
Executive Summary For AI software stocks to move off the bottom, it is not enough to ask whether share prices have already fallen far enough. The more important question is whether the market has seen enough evidence that these companies can still create monetizable value in the AI era. This article argues that the process requires two resets and one valuation threshold. The first is a value reset. Software companies need to prove that they
From Windows RT to RTX Spark: Has Agentic AI Changed the Conditions for Windows on Arm to Succeed?
Executive Summary Windows on Arm has drawn attention several times, yet it has struggled to achieve a meaningful breakthrough. The challenge is not limited to processor performance. It also reflects decades of accumulated compatibility requirements across Windows software, drivers, and enterprise systems. The Chromebook experience further shows that a lighter compatibility burden does not automatically turn Arm’s power-efficiency advantage into market share. Consumers and businesses do not buy an architecture alone. They evaluate the full
Beyond the Traditional Hardware Framework: AI Infrastructure Is Forming a New System Cycle
Executive Summary This AI infrastructure cycle may not be fully understood through the traditional framework of technology hardware cycles. First, although the market keeps discussing an AI bubble, these early bubble warnings may actually make suppliers more disciplined about capital spending and delay the point at which supply becomes excessive. Second, KV cache is changing the architecture of AI infrastructure. AI infrastructure is no longer only a question of supply and demand for individual hardware
In the AI Era, the Market Is Reassessing Software Companies
Executive Summary For some time, software companies have continued to report solid, and sometimes better than expected, earnings results. Yet the market response has remained relatively cautious. This gap may not reflect a problem with company performance. It may reflect a broader reassessment of software companies in the AI era. In the past, software companies could often earn higher valuations through subscription models and predictable growth. In the AI era, the market is beginning to
The Expansion Logic of AI Infrastructure Is Changing
Executive Summary Several recent signals that appear unrelated at first glance may in fact point to a shift in how decisions around AI infrastructure are being made. Adjustments to the expansion pace of the Abilene data center by OpenAI and Oracle, together with Meta’s description of its in-house AI chip roadmap for MTIA, suggest that companies are facing the same underlying question. As model development, chip generations, and infrastructure construction cycles become increasingly out of
A Second Path Beyond the GPU? Architectural Thinking Behind NVIDIA’s Licensing Agreement with Groq
Executive Summary NVIDIA’s licensing agreement with Groq is worth watching not only because the technology itself is extreme, but because it may signal that AI compute architecture is being reconsidered. Even after GPUs have become the dominant platform for AI training and inference, NVIDIA is still willing to engage seriously with an execution model that runs almost counter to the mainstream path. That suggests the demands of the inference era may be making determinism important
CPU as an AI Pillar, Is Arm Approaching a Structural Inflection?
Note (March 2026): I wrote this piece before Arm officially unveiled its own data center CPU. That does not make the original argument irrelevant, but it does change the context in an important way. I am keeping the article largely as it is because the framework still helps explain what to watch. What has changed is that some of the questions discussed here are no longer purely hypothetical. They can now be read
When Grace CPU Reaches Its First Large-Scale Deployment: This Is Not Just a CPU Story but Also a Shift in Data Center Structure
Executive Summary The first large-scale deployment of the Grace CPU may appear, at the surface level, to be a routine update on product and partnership progress. Within a broader industry context, however, this development may carry structural implications that extend beyond a single product milestone. This article examines the signals embedded in Grace CPU’s large-scale deployment from the perspectives of market positioning, data center architectural evolution, and hyperscaler strategy. These signals include NVIDIA’s changing role
AI Is Reshaping the Cost Structure of the Software Industry
Executive Summary From Microsoft to Google, senior executives have increasingly centered their earnings discussions on token efficiency, inference costs, and overall system utilization. This shift in language points to a deeper structural change. As software usage itself begins to incur meaningful costs, the long-held SaaS assumption that higher usage naturally leads to higher margins no longer holds universally. For software companies that lack scale, bargaining power over compute resources, or structural cost advantages, heavy users
In the Age of AI Inference, a Narrative Shift Is Taking Shape
Executive Summary The rapid growth of generative AI has led the market, over the past two years, to focus on memory supply and storage capacity. As AI systems move decisively into an inference-driven phase, however, the fundamental bottlenecks facing infrastructure are beginning to shift. In inference environments, system costs are no longer determined primarily by model size or total data volume. Instead, they are shaped by how contextual states persist during computation. When large volumes