The discussions taking place in some investment circles at the moment have a certain feel to them. Not quite panic. It’s more akin to deliberate patience, the kind that results from having seen this before and experienced enough market cycles to understand the current situation even when the headlines are shouting otherwise. The AI stocks that characterized the past two years’ excitement are now declining.
Some people are really depressed. Oracle has been reduced by half. Since its peak, Microsoft has lost more than 20%. However, the atmosphere is not one of departure in the offices and video conferences where significant long-term funds are being reallocated. It has to do with accumulation. Despite the correction, investors who were forecasting the next tech boom are not. As a result, they are forecasting it.
| Entity / Concept | Detail | Significance |
|---|---|---|
| Gartner Hype Cycle | 5-stage framework: Innovation Trigger → Peak of Inflated Expectations → Trough of Disillusionment → Slope of Enlightenment → Plateau of Productivity | The framework most serious investors are using to interpret the current AI correction as a buying opportunity, not an exit signal |
| Current AI Stage | Trough of Disillusionment (early 2026) | Microsoft down 20%+ from peak; Broadcom down 10%+; Oracle shares halved. NBER survey: 80%+ of 6,000 CEOs/CFOs report AI has had no net-positive impact on employee productivity |
| Bill Gurley (VC Investor) | General Partner, Benchmark Capital — longtime Silicon Valley venture investor | Predicts AI boom will see a “reset” that grounds valuations; expects SaaS stocks to fall to compelling levels before recovery |
| Oracle (NASDAQ: ORCL) | Database and cloud infrastructure company pivoting to AI market | AI infrastructure revenue projected to grow from $18B (2026) to $144B by 2030; stock halved from peak, creating potential entry point |
| Alphabet / Google (GOOGL) | Search, cloud, and AI company; Google Cloud fastest-growing segment | Gemini gaining most ground on ChatGPT; Google Cloud outgrowing AWS and Azure per Synergy Research; quantum computing optionality |
| Broadcom (NASDAQ: AVGO) | Semiconductor company; leader in custom AI ASICs (Application-Specific Integrated Circuits) | Projects $100B+ in AI chip revenue in fiscal 2027 (5× fiscal 2025); building custom chips for Alphabet, Meta, OpenAI |
| AMD (NASDAQ: AMD) | GPU and CPU maker; No. 2 behind Nvidia in GPUs; market leader in data center CPUs | 6 GW GPU purchase commitments from OpenAI and Meta; agentic AI driving CPU demand; both customers hold AMD warrants, incentivizing support |
| Micron Technology (NASDAQ: MU) | Memory chip maker; produces High Bandwidth Memory (HBM) essential for AI GPU performance | Signed first-ever 5-year strategic supply deal; trades at under 4.5× forward P/E on FY2027 estimates; historically cyclical but becoming less so |
| Watch List Names | Recursion Pharmaceuticals (AI drug discovery); UiPath (workflow automation) | Lower-profile AI applications with specific, demonstrable enterprise use cases — not dependent on mass consumer AI adoption |
| Historical Parallel | Dot-com bust (2000–2002) | Most companies failed; survivors — Amazon, Google — became cornerstones of modern internet. Investors who bought the trough saw generational returns |
| Reference | Motley Fool — Trough of Disillusionment Analysis | |
The Gartner Hype Cycle, a model that technology research firm Gartner formalized decades ago to describe how new technologies move through public consciousness and capital markets, is the framework that the majority of them are silently working from. There are five stages: an innovation trigger, a peak of exaggerated expectations, a slope of enlightenment, a trough of disillusionment, and a productivity plateau. The majority of seasoned investors have come across this framework at some point, so it’s not very contentious. However, when markets are at their lowest point and the temptation to conclude that the technology was never real is at its strongest, its predictive power is often undervalued.
It is easy to find evidence that AI is currently at its lowest point. More than 80% of respondents to a National Bureau of Economic Research survey, which gathered data from over 6,000 chief executives and chief financial officers, stated that artificial intelligence was not improving worker productivity. Sitting with that number is worthwhile.
These individuals are in charge of the businesses, purchasing subscriptions, approving IT budgets, and reporting to boards regarding ROI. Eighty percent of them claim that the technology isn’t working, which has implications for the stocks of all investors. The repricing is explained. It doesn’t explain the conclusion that some investors are coming to, which is that a technology that doesn’t increase productivity now won’t do so tomorrow. The next item on Gartner’s list is disregarded in that conclusion.
The AI boom will see a “reset” that grounds valuations and reframes spending, according to venture investor Bill Gurley, who has experienced enough cycles to have opinions worth listening to. He is correct that there was a need for a reset, and it may already be taking place. However, a reset in the Gartner model is a correction that eliminates the least tenable presumptions and leaves the more sustainable companies behind, rather than a collapse.
The clear historical parallel is the dot-com bust of 2000, which astute investors consistently revisit. The majority of those businesses have vanished. Webvan, Pets.com, and all the other names that generated revenue based on metrics that proved to be ornamental. Amazon and Google, two of the survivors, went on to shape the contemporary economy. While the rest of the industry was going dark, they were purchasing servers during the downturn.
Investors who forecast the next tech boom are placing specific wagers rather than broad ones. Oracle, a database company for the majority of its existence before making a conscious shift to AI infrastructure, predicts that its AI revenue will increase from eighteen billion dollars this year to one hundred and forty-four billion dollars by 2030. If that amount comes to pass even close to its stated amount, it would be akin to a total business transformation.
The other name that keeps coming up is Alphabet; this isn’t due to ChatGPT’s continued dominance in the AI chatbot race, but rather to Google Cloud’s enterprise penetration, Google Gemini’s victories over market-leading AI products, and the fact that Alphabet is best positioned to enter the quantum computing space when that technology becomes commercially viable. The AI play isn’t the most obvious. That might be the exact reason it’s worth seeing.
Another reason Broadcom is attracting attention is that it manufactures the application-specific integrated circuits, or custom AI chips, that hyperscalers like Alphabet, Meta, and OpenAI are increasingly selecting over commercial GPU alternatives. In fiscal 2027, Broadcom anticipates earning over $100 billion from AI chips. In the meantime, Meta and OpenAI have committed to buying six gigawatt GPUs from AMD, and both clients have AMD warrants that provide them with financial incentives to support AMD’s growth.
Such structural alignment is uncommon and noteworthy. Additionally, Micron Technology, which is currently trading at less than five times forward earnings on 2027 projections, provides the high-bandwidth memory that enables all AI chips to function. Longer-term supply agreements with clients who require consistent access are gradually smoothing out the company’s historically cyclical nature.
The duration of the trough and whether the slope of enlightenment emerges in the clear, narrative arc that the framework suggests are still unknown. Markets don’t always adhere to models. However, there’s a sense that the shape of this moment is familiar to those who have witnessed it before, as they watch the investors who got dot-com right accumulate positions in these particular names. The technology is here to stay.
It will find applications in cybersecurity, drug development, workflow automation, and other areas that don’t make for compelling customer stories but produce steady business income. Investors who have already positioned themselves for that AI version are not waiting for favorable headlines. They had long since given up waiting.
