Morgan Stanley Assesses That the AI Market Has Reached an Inflection Point: Moving from Computing Power Construction to Industrial Diffusion, with a Barbell Strategy Becoming the Main Allocation Approach

Morgan Stanley Assesses That the AI Market Has Reached an Inflection Point: Moving from Computing Power Construction to Industrial Diffusion, with a Barbell Strategy Becoming the Main Allocation Approach Morgan Stanley has released a thematic research report titled "Decoding the Rhythm of AI Transformation: From Computing Power Construction to Industrial Diffusion." Based on its sixth-round AI stock mapping database covering approximately 3,600 stocks globally, it presents a core judgment: AI investment has moved beyond the stage of simply betting on upstream computing power enabler targets and entered a new era of "barbell-style" dual-track allocation, requiring investors to both capture structural opportunities brought by computing power bottlenecks and increase allocation weight to companies adopting AI applications. Morgan Stanley states that its core research theme for 2026 is technology diffusion. In this industrial cycle currently in transition, it expects a barbell-style excess return pattern, with core infrastructure enabler companies and early-stage software vendors and AI application adoption companies generating gains simultaneously. The typical pattern of technology cycles is that semiconductors are the first to achieve excess returns, followed by the infrastructure sector, and then software and services. The bank judges that the investment logic of the AI industry is entering a new phase: shifting from trades highly concentrated in enabler companies to a broader barbell-style opportunity pool, covering both select enabler targets and emerging AI application adoption companies. Since the start of this AI cycle, the semiconductor and infrastructure sectors have recorded 500-700 basis points of excess returns relative to the S&P 500. We believe it is now time to begin increasing allocation exposure to early-stage software enabler companies and AI application adoption companies. At the same time, however, the massive AI capital expenditure construction wave, combined with key bottleneck factors such as electricity, will extend the upside window for certain high-quality targets. Therefore, this computer cycle will not replicate the simple linear rotation of market-leading sectors seen in past cycles. The bank notes that this AI cycle has unique characteristics, with labor, electricity, and policy regulation constituting three core constraints. The report estimates that from 2026 to 2028, global data centers will face a 57GW electricity gap. Combined with practical obstacles such as local approvals and labor shortages, the pace of computing power supply expansion will be significantly suppressed. The computing power shortage will persist for years, and there will not be the simple, clear sector handoff rotation seen in history. Hyperscaler capital expenditure growth is expected to decline from 93% in 2026 to 14% in 2028. The expectation that hardware capital expenditure prosperity has peaked is gradually being priced in by the market, but the prosperity of certain bottleneck segments will continue. Data shows that AI enabler companies (upstream suppliers such as chips, computing power, and hardware) have doubled their expected EPS over the past two years and continue to maintain strong earnings momentum. Meanwhile, AI adoption companiescompanies across all industries that apply AI technology to their own businesses to achieve cost reduction and efficiency improvementare expected to see cumulative EPS growth of approximately 70% over two years in the next 12 months, with an earnings inflection point already apparent. Market consensus expects that adoption companies with high AI substantive impact will see EBIT margin expansion of 460 basis points in 2025-2026, nearly double that of the MSCI ACWI index. However, the market has not yet fully priced in the long-term dividends brought by AI production efficiency, and there is room for upward revision in forward earnings forecasts. On valuation, the forward P/E of adoption companies with high AI impact has fallen back to 18x, compared to 22x for enabler companies. After valuation digestion, the risk-reward ratio has improved significantly. The bank's proposed barbell investment strategy is divided into two ends. One end continues to hold bottleneck assets: prioritizing power-related segments, including energized data center service providers, energy storage, grid equipment, and new energy; while also selecting individual stocks within semiconductors and infrastructure, focusing on second-layer supply chain bottlenecks beyond power. The other end expands new allocations: increasing allocation to early-stage software enabler companies, with priority order being infrastructure software, cybersecurity, and then select application software; positioning in AI adoption companies that have already delivered quantifiable business returns, with screening focused on two key indicatorsfirst, the substantive impact of AI on the company's business, and second, corporate pricing power. Only companies that can retain the cost-reduction dividends of AI can truly deliver on profits. IT services belong to late-cycle beneficiary segments and are not yet suitable for focused allocation at this stage. From a global perspective, the AI trajectory across markets is clearly diverging. In North America, AI dividends are diffusing from hardware to software and the real economy; in Asia-Pacific, returns remain concentrated in the upstream supply chain, but an increasing number of companies are disclosing AI-driven revenue increments in their financial reports; in Europe, AI opportunities are more concentrated in application deployment in traditional industries. The report particularly emphasizes that investment cannot merely chase "AI concepts"the substantive impact of AI on a company's investment logic is the key to success or failure. Targets where AI is a core logic enjoy enormous excess returns compared to those where it has only a secondary impact; conversely, companies where AI constitutes a core business threat will significantly underperform peers facing only moderate disruption. At the same time, there are reverse risks to the market. If computing power demand continues to exceed expectations and labor and electricity bottlenecks persist, the prosperity duration of the semiconductor and infrastructure sectors may far exceed market baseline expectations. Historically, Wall Street has repeatedly underestimated the development space of the technology industry. Overall, the AI market has moved from pure "stacking computing power" thematic speculation toward a new stage of validating commercial ROI. Balancing upstream bottleneck assets with application targets that deliver on deployment has become the allocation approach better suited to the current environment.