Citigroup says global wafer fabrication equipment spending could climb to $250 billion by 2028 in a bullish scenario, as artificial intelligence and high-performance computing continue to reshape semiconductor capacity plans. The projection points to one of the largest potential equipment cycles in the chip industry’s history, but it rests on a narrow group of companies and a demanding set of assumptions.
The main test will come from Taiwan Semiconductor Manufacturing Co., Samsung Electronics and Intel, whose capital spending plans are expected to determine whether the current upcycle extends beyond 2026. Together, the three chipmakers are projected to account for about 55% of global wafer fabrication equipment spending in 2025, making their guidance for 2027 and 2028 crucial for suppliers of lithography tools, etching systems, deposition equipment, inspection machines and advanced packaging technology.
Citigroup’s bullish model assumes global wafer fabrication equipment spending will reach $145 billion in 2026, rise to $200 billion in 2027 and then hit $250 billion in 2028. That path would mark a major expansion from prior semiconductor cycles, reflecting the growing role of AI processors, high-bandwidth memory, advanced logic, and packaging-intensive computing systems.
The near-term focus is now shifting to earnings season. TSMC is scheduled to report results on July 16, Intel on July 23, and Samsung will hold its quarterly call on July 30 after issuing preliminary guidance earlier in the month. Traders will be watching not only 2025 and 2026 figures, but also any language that points to capacity buildouts in 2027 and beyond.
TSMC remains the central swing factor
TSMC is the most important company in the forecast because of its dominant position in advanced semiconductor manufacturing. Citigroup’s projection assumes that TSMC’s 2027 capital expenditure could reach $75 billion, up 36% from the prior year and far above current market expectations.
That would be an aggressive step-up, even for a company already spending heavily to support advanced process nodes and packaging. TSMC said during its April earnings call that 2026 capital spending was expected to fall in a range of $52 billion to $56 billion. That guidance suggested continued allocation toward leading-edge manufacturing technologies and advanced packaging capacity, both of which are essential for AI accelerators and high-performance computing chips.
The challenge is that moving from that 2026 range to a $75 billion figure in 2027 would require sustained demand visibility. Large cloud service providers, chip designers and enterprise customers would need to keep expanding AI infrastructure at a rapid pace. At the same time, TSMC would need to see enough confidence in end-demand to justify new tools, facilities and packaging capacity.
Advanced packaging remains a key constraint. AI processors increasingly rely on complex packaging technologies to connect logic chips with high-bandwidth memory. If packaging bottlenecks ease, tool demand could accelerate. If they persist, equipment orders may be spread out over a longer period.
TSMC’s comments on utilization, customer commitments and capacity tightness will therefore carry substantial weight. Traders will be looking for signs that demand is broadening beyond the largest AI chip programs and into a wider range of custom silicon, networking processors, server components and edge AI devices.
Samsung’s memory and logic plans are under review
Samsung is another major factor in the bullish equipment scenario. The company’s contribution is expected to come from high-bandwidth memory, DRAM and advanced logic, all areas tied directly or indirectly to AI infrastructure.
The company has outlined a long-term semiconductor spending plan that exceeds 2,000 trillion Korean won over more than a decade. That figure signals the scale of Samsung’s ambitions, but it does not automatically translate into near-term equipment purchases. The timing and composition of actual orders will depend on demand, pricing, technology yields, and the pace at which Samsung expands capacity for memory and foundry services.
High-bandwidth memory is particularly important. AI servers rely heavily on HBM because it allows processors to access large amounts of data quickly and efficiently. Demand for these memory stacks has increased sharply as AI model training and inference workloads have expanded. If Samsung can gain further traction in HBM supply, its equipment needs could rise meaningfully.
DRAM is another key area. The memory market has historically moved in cycles, with periods of tight supply followed by oversupply. AI demand has improved the outlook for premium memory products, but broader DRAM conditions still matter. If conventional memory demand weakens, Samsung may delay some equipment spending even if AI-related products remain strong.
Samsung’s advanced logic strategy is also important for the global wafer equipment market. A stronger push into next-generation foundry manufacturing would require substantial spending on advanced production tools. However, the company faces intense competition from TSMC, while customer adoption and yield progress remain central questions.
Intel’s foundry execution could lift or limit the cycle
Intel provides a different type of upside risk. The company’s spending outlook for 2027 and 2028 depends heavily on whether its foundry transformation gains clear commercial momentum.
Intel is trying to become a larger contract chip manufacturer while also supporting its internal product roadmap. The success of that strategy will depend on process technology execution, external customer wins and confidence in manufacturing timelines.
Citigroup’s scenario assumes potential upside if Intel makes progress with its 18A process and secures design wins at the 14A node. These milestones are important because they would show whether Intel can attract outside customers to its most advanced manufacturing platforms.
If Intel qualifies 18A successfully and demonstrates strong yields, confidence in the foundry plan could improve. That would likely support additional equipment spending. If design wins at 14A also materialize, the company may need to invest further in capacity, advanced tools and related infrastructure.
The reverse is also true. If customer traction falls short, if qualification timelines stretch, or if demand visibility remains limited, Intel may be more cautious with capital spending. That would reduce the likelihood of the bullish global equipment scenario.
For equipment suppliers, Intel’s path matters because the company has historically been a major buyer of advanced manufacturing tools. A stronger Intel foundry business would add another source of demand for leading-edge wafer equipment, especially in the United States and Europe, where governments are pushing for more domestic semiconductor capacity.
Micron adds strength, but the three giants still dominate
Micron has also raised its capital expenditure plans, adding support to the broader semiconductor equipment outlook. The memory producer has guided for about $27 billion in spending in fiscal 2026 and has indicated that 2027 spending could exceed $40 billion.
That planned increase reflects strong demand for HBM and DRAM used in AI servers. Like Samsung, Micron is benefiting from the need for memory-intensive computing systems. AI accelerators require large volumes of advanced memory to operate efficiently, and that trend has improved the strategic importance of memory producers.
Still, Micron alone cannot determine the global wafer equipment cycle. The overall direction will remain heavily influenced by TSMC, Samsung and Intel because of their combined scale. Those companies control a large share of leading-edge logic, foundry and memory-related expansion decisions.
Micron’s spending does, however, signal that the AI-related memory boom is not limited to a single supplier. It also shows that the AI infrastructure buildout is affecting multiple layers of the chip supply chain, from logic processors to memory stacks and packaging.
AI demand is driving equipment intensity
The most important force behind the bullish forecast is the rising equipment intensity of advanced computing. Modern AI chips require increasingly complex manufacturing processes. Leading-edge nodes need more advanced lithography, tighter process control and additional inspection steps. Packaging technology has also become more important as chip designers combine processors, memory and interconnects into increasingly dense systems.
This means that each new generation of AI hardware can require more equipment spending than previous computing cycles. The industry is not simply adding more capacity; it is adding more complex capacity.
Demand from cloud platforms remains the strongest visible driver. Large data center operators are building clusters designed for AI training and inference, while enterprises are beginning to adopt AI tools that require more specialized computing. Networking, storage and power management components are also becoming more important as AI server deployments scale.
High-performance computing demand is another support. Scientific computing, advanced simulation, defense applications, autonomous systems and data analytics all require powerful chips. These workloads may not move at the same pace as generative AI, but they contribute to the long-term need for advanced semiconductors.
Risks could slow the path to $250 billion
The $250 billion figure is possible only if capital spending keeps rising after 2026. That is not guaranteed.
Several risks could interrupt the timeline. Equipment delivery delays remain a concern, especially for the most advanced manufacturing tools. Building semiconductor capacity takes time, and bottlenecks in one part of the tool chain can slow entire fab projects.
Macroeconomic conditions also matter. If global growth weakens, if corporate technology spending slows, or if financing conditions tighten, chipmakers may become more cautious. Semiconductor companies often adjust capital expenditure quickly when demand visibility changes.
Capacity utilization is another key risk. If companies build too much capacity too quickly, utilization rates could fall. That would pressure margins and likely lead to slower equipment orders. The semiconductor industry has seen this pattern before, particularly in memory markets.
A softer AI demand environment would be the biggest threat to the bullish scenario. Current spending plans assume that AI infrastructure demand remains strong for several years. If customers reduce orders, delay data center expansion or shift to more efficient chip architectures that require fewer units, equipment demand could slow.
Geopolitical factors also remain relevant. Export controls, government subsidies, regional manufacturing policies and supply-chain restrictions can all affect where and when factories are built. These factors may support spending in some regions while delaying projects in others.
Broader chip forecasts point to a larger ecosystem
A separate semiconductor sector forecast cited by Bank of America recently projected that the global chip market could reach $1.3 trillion by the end of the current year. Such projections reflect the growing value of semiconductors across AI, automotive systems, industrial automation, consumer electronics, data centers and communications infrastructure.
However, a larger semiconductor market does not automatically guarantee an immediate flood of specialized processing units into global supply chains. Capacity expansion must still pass through design cycles, equipment delivery schedules, factory construction, yield qualification and customer adoption.
The same caution applies to downstream markets such as custom validation hardware for digital ledger networks. Forecasts have suggested that specialized validation hardware could reach a valuation of about $12.46 billion before year-end, but that market remains dependent on token economics, power costs, hardware efficiency and regulatory conditions.
Cheaper components may reduce the cost of building large validation rigs over time, but the relationship is not automatic. Hardware prices can fall because of stronger supply, weaker demand, technological substitution or secondary-market liquidation. Each cause has different implications.
For traders exposed to digital assets, semiconductor supply trends may become a useful signal, but they should not be treated as a simple hedge or a guaranteed source of protection. Equity in hardware suppliers can move with earnings, margins, customer concentration, export controls and broader stock-market conditions. Digital tokens can move for entirely different reasons, including liquidity, protocol changes, regulatory headlines and risk appetite.
Used hardware markets may feel the impact
The growth of dedicated machine-learning chips could also affect secondary hardware markets. Some projections suggest the machine-learning chip industry could expand sharply in the near term, potentially reaching $125 billion in value within months if demand continues at its current pace.
If corporate server farms replace older graphics processing units with newer AI accelerators, more used equipment could appear on secondary markets. That may put downward pressure on prices for certain older GPUs, servers, cooling units and related components.
Still, timing is uncertain. Large data center operators do not always sell older equipment immediately. Some hardware is repurposed internally for less demanding workloads. Other systems are held for redundancy, regional deployments or resale at a later date.
For personal node operators and smaller compute buyers, waiting for clearer secondary-market pricing may be sensible, but a precise price bottom is difficult to forecast. Hardware availability depends on corporate refresh cycles, warranty terms, energy efficiency, software compatibility and logistics.
Cooling systems are another area to watch. Higher factory output could shorten delivery times for equipment used in dense computing centers, but demand is also rising. AI servers generate significant heat and require more advanced cooling infrastructure, including liquid cooling in some deployments. If AI data center construction accelerates faster than component output, delivery times may remain tight.
Distributed storage tokens remain a speculative link
Some market participants also connect semiconductor production trends with distributed storage networks and other utility-token ecosystems. The argument is that greater availability of computing and storage hardware can lower operating costs and expand network capacity.
That relationship exists in theory, but token prices do not necessarily rise whenever hardware production increases. Distributed storage networks depend on actual usage, developer activity, data demand, emissions schedules, governance decisions and competition from traditional cloud providers.
A spike in global machinery production may improve the economics for node operators, but it does not guarantee higher token prices. Traders should separate the infrastructure case from the market-price case. Lower hardware costs can help network supply, while token valuations depend on whether demand grows enough to absorb that supply.
Earnings guidance becomes the next major signal
The central question for the semiconductor equipment cycle is whether 2026 represents a peak or a stepping stone toward even larger spending in 2027 and 2028.
Citigroup’s bullish $250 billion wafer fabrication equipment forecast requires several developments to happen together. AI chip demand must remain strong. TSMC must continue expanding leading-edge and packaging capacity. Samsung needs to convert long-term semiconductor ambitions into visible equipment orders. Intel must show that its foundry roadmap can attract customers and justify higher spending. Memory producers such as Micron must keep seeing durable demand for HBM and advanced DRAM.
The coming earnings reports will not settle every question, but they should provide important clues. Traders will focus on capital spending ranges, comments on AI demand, packaging bottlenecks, process-node roadmaps, customer commitments and equipment lead times.
If the three largest chipmakers signal confidence in 2027 and 2028 demand, the equipment upcycle could extend well beyond current expectations. If they remain cautious, the $250 billion scenario may look more like an upper-bound case than a base-case forecast.
Curious how AI demand shapes markets like chips and crypto? Explore our in-depth guide to digital assets next.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
