Intel Commits to 14A Mass Production in 2028, but TSMC Sets the Test

Intel increased quarterly revenue by 25% and committed to high-volume 14A production in 2028, giving the Tom Hardware readership a striking turnaround story. The stronger results arrived alongside recovering margins, record server growth, and rising demand for Intel’s manufacturing capacity. Yet Intel Foundry still recorded a substantial operating loss, while external 14A customer commitments remain undisclosed.

That tension matters more than the headline growth rate. Intel is promising to finance and execute another advanced manufacturing transition while its current foundry operation remains deeply unprofitable. Its 14A schedule also puts the company into a direct timing contest with TSMC, which plans to begin A14 volume production in 2028.

Intel therefore has two different achievements to prove. It must convert better product demand into durable earnings, then turn 14A from an internal manufacturing program into a credible commercial foundry service. The first task is helping fund the second, but one strong quarter does not complete either job.

Intel’s Strong Quarter Changes the Starting Point

Intel enters the 14A investment cycle with stronger demand, better margins, and more operating cash than it had one year earlier.

Intel reported second-quarter 2026 revenue of $16.1 billion, a 25% increase from the same period in 2025. Revenue landed $1.8 billion above the midpoint of management’s earlier guidance. Intel described the result as its strongest revenue growth in more than 15 years.

The company also reported non-GAAP earnings per share of $0.42. That figure exceeded Intel’s prior guidance and indicated that its underlying operations were profitable after selected adjustments. The quarterly results also showed $7 billion in operating cash flow.

Intel’s reported GAAP loss complicates that picture. The company recorded an $11 billion GAAP net loss, including a $13.619 billion mark-to-market charge tied to escrowed shares. Those shares relate to Intel’s Secure Enclave agreement with the United States government.

That accounting charge does not describe daily factory performance, but it cannot simply be ignored. GAAP results capture obligations and changes that adjusted figures remove. Readers should therefore treat Intel’s non-GAAP profit as evidence of operational improvement, not proof that every financial problem has disappeared.

Gross margin offers a clearer operating signal. Intel’s GAAP gross margin reached 40.1%, compared with 27.5% in the second quarter of 2025. Gross margin measures the share of revenue remaining after direct production costs, so the increase suggests better factory economics and product mix.

The revenue gains also came from identifiable businesses. Intel’s Client Computing and Physical AI Group generated $8.9 billion, up 13% year over year. Its Data Center and AI business reached $6.3 billion, representing 59% annual growth.

Server demand supplied the clearest source of momentum. Chief executive Lip-Bu Tan said Intel recorded its strongest year-over-year server growth on record. He also described Xeon 6 as one of the fastest-ramping products in the company’s history.

Purpose-built silicon, which includes chips designed for narrower customer workloads, also contributed. Intel said revenue from that business rose roughly 20% from the previous quarter and nearly tripled from one year earlier. These gains give Intel exposure beyond conventional PC and server processors.

The client result deserves more caution. Intel said the increase did not come primarily from higher unit shipments. Chief financial officer Dave Zinsner attributed much of the performance to higher average selling prices, product mix, and cost inflation passed to customers.

Higher prices can support revenue and margins when supply is tight. They are less persuasive as evidence of expanding market demand if unit sales remain flat. Intel’s client business will eventually need either higher volumes or a sustained premium mix to preserve this growth rate.

Even so, the quarter changes Intel’s immediate position. The company is no longer presenting 14A while every major operating indicator points downward. It can now connect that manufacturing commitment to rising internal demand, improving yields, and stronger cash generation.

That connection explains why management committed to the 2028 ramp now. Demand has increased the potential value of leading-edge capacity, while better operations have improved Intel’s ability to fund it.

Tom Hardware Finds the Real Story in Intel 14A

The important 14A announcement is not a laboratory milestone. Intel has authorized a high-volume ramp before announcing a committed external manufacturing customer.

Intel says 14A remains on track for risk production during the second half of 2027. Risk production is the early manufacturing stage used to validate a process and customer design before full commercial output. The company decided during the second quarter to commit fully to a high-volume ramp in 2028.

Tan connected that decision to two sources of demand. Intel sees greater interest from its external foundry prospects, and its internal product teams expect to need more 14A capacity. The Tom Hardware analysis emphasizes that Intel’s own products will lead the initial ramp.

This distinction matters. High-volume manufacturing sounds like broad commercial availability, but Intel can reach that milestone by producing its own processors. An outside chip designer still needs validated design tools, competitive economics, sufficient capacity, and confidence in Intel’s delivery schedule.

Intel has not publicly identified a firm external customer committed to manufacturing a high-volume 14A product. The company has previously discussed prospective customers and external engagement. Those discussions do not yet equal a disclosed production contract.

The 2028 commitment nevertheless carries financial weight. A leading process requires equipment orders, factory preparation, process development, and trained manufacturing teams years before customers receive finished chips. Intel cannot wait until every sales agreement is complete before making those investments.

Its internal products provide a baseline load for the factories. Intel can use its processors to refine yields, increase wafer volume, and identify manufacturing problems. Yield is the share of usable chips produced from a wafer, and it strongly affects the cost of each sellable component.

This integrated model gives Intel an advantage that pure-play foundries lack. Intel designs products and manufactures silicon, so it can become its own first customer. Internal demand can help fill an early production line when outside customers remain cautious.

The same model creates a credibility problem. Internal volume does not prove that an independent chip company wants Intel as a supplier. It also does not show that Intel can provide the predictable tools, support, confidentiality, and schedules expected from a commercial foundry.

That gap is the real conflict behind the announcement. Intel can declare 14A a manufacturing success after shipping its own products. The market will judge Intel Foundry more strictly, using external revenue, customer tape-outs, yields, margins, and repeat orders.

A tape-out is the point when a completed chip design is sent for manufacturing preparation. It represents a more concrete commitment than general technical evaluation. Intel has not disclosed a comparable portfolio of committed 14A tape-outs from outside clients.

The timing also remains broad. A high-volume ramp late in 2028 would produce a different competitive result from a ramp early that year. Products built during a late ramp might not reach customers until 2029.

Development-fab output can further blur the milestone. Intel normally begins advanced-node manufacturing in Oregon, where early volumes can be lower than output from a large commercial factory. The label matters less than wafer starts, usable yields, and customer shipments.

The Tom Hardware framing is therefore useful because it keeps the financial and manufacturing stories connected. Intel’s stronger product demand supports 14A investment. However, the 14A commitment also places new spending and execution pressure on a foundry operation that has not reached profitability.

The 2028 Race Puts Intel Against TSMC

Intel’s primary test is whether 14A can become a credible alternative to TSMC’s A14, not whether both companies can announce production during 2028.

TSMC has scheduled its A14 process for volume production in 2028. The company says A14 will use its second-generation nanosheet transistor structure, which controls electrical current through stacked horizontal channels. That design targets higher performance and lower energy use.

TSMC projects a 10% to 15% speed improvement at the same power compared with N2. Alternatively, it expects a 25% to 30% power reduction at the same speed. The company also forecasts a logic-density gain close to 20%, according to its A14 update.

These remain company projections rather than independent measurements from shipping products. Still, TSMC has a commercial advantage that Intel cannot match through technical specifications alone. Its business already revolves around manufacturing chips for outside designers at enormous scale.

TSMC says it sees high customer interest in A14 across smartphones and high-performance computing. It has also said customers have completed A14 tape-outs. That indicates some designs have moved beyond exploratory conversations.

Intel’s route begins from a different place. It can use internal processors to generate 14A volume, then use the resulting yield and reliability data to attract outside customers. The strategy reduces the risk of opening an empty production line.

However, potential foundry clients must choose suppliers long before finished products ship. They need process design kits, which are collections of models, rules, and tools used to build chips for a particular process. They also need reliable libraries, packaging options, and manufacturing forecasts.

Moving a major design between foundries is not a simple substitution. Each process has different design rules, performance characteristics, and supporting technology. A customer that commits deeply to TSMC’s A14 cannot casually move the same design to Intel 14A after a delay.

That makes schedule credibility commercially decisive. If Intel delivers every technical milestone but customers remain uncertain about dates, they can still choose TSMC. Predictability can outweigh a narrow performance advantage because delayed chips lose market windows.

Intel also needs to offer a complete manufacturing service. Advanced chips increasingly depend on packaging that combines processors, accelerators, memory, and communication components. A strong transistor process without sufficient packaging capacity may not satisfy large AI customers.

TSMC has spent years building relationships with companies that compete against Intel’s product groups. Intel must persuade those businesses that its foundry organization can protect their information and prioritize their schedules. Organizational separation and customer service are therefore part of the technical contest.

The competition does not require Intel to replace TSMC. Even a modest share of advanced external manufacturing could create a meaningful business if Intel improves utilization and margins. Large chip designers also have reasons to seek a second source outside Taiwan.

Supply-chain diversification supports Intel’s pitch, particularly for United States manufacturing. Governments and customers want more geographic options for strategically important chips. Intel’s domestic factory footprint gives it relevance beyond benchmark results.

Yet diversification does not excuse weak economics or late delivery. Customers will not move their highest-value designs solely to make a geopolitical statement. Intel must combine location, technology, capacity, and reliable execution into a competitive offer.

This is why the shared 2028 target should not be read as a tie. TSMC enters that year with established external customers and a proven foundry model. Intel enters with internal products, improving factories, and a chance to earn external trust.

Better Yields Have Not Fixed Intel Foundry

Intel Foundry is improving, but its losses and limited external revenue show how far the business remains from validating the 14A strategy.

Intel Foundry recorded $5.8 billion in second-quarter revenue, up 31% from one year earlier. Much of that figure reflects manufacturing work performed for Intel’s own product groups. External foundry revenue was only $293 million.

The segment reported a $2.1 billion operating loss. That was better than the $2.4 billion loss recorded in the preceding quarter and the $3.2 billion loss one year earlier. Progress is visible, but the operation still consumes substantial capital.

Zinsner said higher yields improved production cycle times and reduced wafer costs across Intel 4, Intel 3, and 18A. Factory scale also helped. These are the operational improvements Intel must repeat before 14A reaches risk production.

Cycle time measures how long a wafer takes to move through the manufacturing process. Shorter cycles can release capacity, reduce inventory, and help engineers identify defects more quickly. Better yields then spread factory costs across more sellable chips.

Intel 18A is the immediate proving ground. Its ramp can show whether Intel has restored manufacturing discipline before the company attempts another transition. A stable 18A operation would support management’s claim that 14A starts from a stronger foundation.

Conversely, unresolved 18A problems would place more risk on the 14A schedule. Advanced processes contain thousands of tightly controlled steps, so development progress does not guarantee economical mass production. Small yield differences can determine whether a node earns money.

The foundry result also exposes a scale issue. Internal transfers can increase reported segment revenue, but they do not provide the same market validation as an external customer payment. Investors need both measures to understand the business.

Intel’s product groups benefit when internal manufacturing improves. Lower wafer costs and better availability can raise corporate margins even without large external foundry sales. That benefit makes the factories strategically useful before Intel Foundry becomes independently profitable.

The commercial ambition goes further. Intel wants outside customers to use its factories, spreading fixed costs across more wafers and creating a separate revenue stream. Reaching that goal requires external sales to grow much faster than their current level.

Intel has raised its capital-spending outlook for 2026 and 2027 as demand exceeds available supply. That response is rational if the demand persists and new equipment increases profitable output. It becomes dangerous if capacity arrives after demand shifts or customers choose another supplier.

Semiconductor spending decisions have long lead times. Factories and specialized tools cannot be added or removed quickly. Intel must therefore make 14A commitments using forecasts that extend beyond the present server boom.

This is where the 25% revenue increase can mislead. The quarter gives Intel more resources and validates current demand. It does not automatically forecast demand during the full 14A production cycle.

Client revenue illustrates the problem. Higher average selling prices supported growth while supply remained constrained. If supply normalizes and pricing weakens, that contribution may not repeat at the same rate.

Data-center demand appears structurally stronger, but Intel competes for spending with AMD processors, Arm-based server chips, and specialized accelerators. AI infrastructure can expand rapidly while changing the mix of components purchased. Growth in total computing demand does not guarantee equal growth for every CPU supplier.

Intel’s 14A case therefore rests on execution across several connected layers. The company must stabilize current nodes, supply competitive internal products, attract external designs, and control capital intensity. Failure at one layer can weaken the economics of the others.

The Headline Numbers Still Need a Stress Test

Intel has presented credible evidence of recovery, but the quarter does not prove that its revenue growth or 14A timetable will endure.

The first uncertainty concerns revenue quality. Intel’s 25% annual increase combines strong server demand, purpose-built silicon growth, and client pricing effects. Those drivers have different levels of durability.

Xeon 6 demand indicates that Intel remains important in data centers. Customers often deploy server platforms for years, which can support continuing processor and service demand. Still, competing architectures can capture new workloads as buyers optimize for energy use and AI performance.

Purpose-built silicon offers another growth path. Custom chips can deepen relationships with large customers and make use of Intel’s design, packaging, and manufacturing capabilities. Yet large customers also have negotiating leverage and can divide work among several suppliers.

The second uncertainty concerns profit measurement. Non-GAAP net income of $2.2 billion shows positive adjusted operations. The $11 billion GAAP loss shows that Intel’s financial structure still contains material complications.

Neither figure tells the complete story alone. Adjusted earnings can clarify operating performance when a large non-cash charge distorts comparisons. GAAP results remain essential because excluded items can still affect shareholders and future flexibility.

The third uncertainty concerns 14A timing. Intel says risk production remains on schedule for the second half of 2027, followed by high-volume production in 2028. That schedule leaves a wide range of possible commercial outcomes.

A ramp early in 2028 with strong yields would strengthen Intel’s competitive position. A limited ramp late in the year could still satisfy the literal guidance while delaying meaningful product availability until 2029. Readers should watch dates and volumes, not just milestone labels.

The fourth uncertainty concerns external customers. Intel has reported encouraging engagement, but it has not publicly named a committed high-volume 14A manufacturing client. An internal product ramp validates technology differently from an external tape-out and production order.

The distinction is particularly important because TSMC already operates as a trusted supplier to many leading chip designers. Intel must overcome switching costs and concerns created by its product competition. A promising process alone does not erase those barriers.

The fifth uncertainty concerns capacity economics. Intel says demand exceeds current supply, encouraging additional investment. Yet new capacity must arrive at the right time and reach satisfactory utilization.

Low utilization raises the cost assigned to every wafer. That can make a technically sound process commercially unattractive. External customers would help fill factories, but Intel must invest before many of those commitments become public.

The official earnings release forecasts third-quarter revenue between $15.8 billion and $16.8 billion. Intel also projects a 42% non-GAAP gross margin and non-GAAP earnings per share of $0.38.

That outlook suggests management expects the business to remain stronger than it was one year earlier. It also gives readers a near-term test of execution. Revenue near the upper end, combined with margin improvement, would support the recovery narrative.

A weaker mix would challenge it. Revenue can remain high while foundry losses, pricing pressure, or unfavorable products hold back profits. The most useful question is not whether Intel grows, but whether each additional unit of growth improves its economics.

The Tom Hardware headline captures a genuine change in direction. Intel is growing again and has committed to 14A production. The skeptical reading is that the hardest evidence, external advanced-node customers and profitable foundry scale, has not arrived.

That is not a reason to dismiss the announcement. It is a reason to separate a funded commitment from a completed turnaround. Intel has moved the burden of proof forward, from survival toward delivery.

Three Signals Will Decide Whether the Turnaround Holds

Intel’s next results, 18A manufacturing performance, and external 14A commitments will determine whether this quarter marks a durable shift.

The first signal is Intel’s third-quarter financial performance. Management expects revenue between $15.8 billion and $16.8 billion, with a 42% non-GAAP gross margin. Results within those ranges would show that second-quarter strength was not an isolated spike.

The composition will matter as much as the total. Investors should compare client units with pricing, track Xeon demand, and examine purpose-built silicon growth. Continued revenue expansion based mainly on constrained supply and higher prices would be less convincing than broad unit demand.

Foundry operating losses deserve equal attention. Another sequential reduction would indicate that improved yields and factory scale are changing the segment’s cost structure. A reversal would suggest that manufacturing progress remains uneven.

The second signal is the 18A ramp. Intel needs to show that higher yields, faster cycle times, and growing output translate into competitive products. Reliable 18A production would strengthen confidence in the company’s ability to prepare 14A.

Intel should also provide enough detail to distinguish technical readiness from profitable manufacturing. Yield percentages may remain confidential, but product availability, factory output, margin trends, and customer shipments can offer indirect evidence.

The third signal is a firm external 14A commitment. Intel previously said prospective customers were evaluating the process, with supplier decisions expected during the second half of 2026. A named customer, completed tape-out, or committed capacity agreement would move 14A beyond internal validation.

The absence of an announcement would not prove that negotiations failed. Foundry contracts can remain confidential, and customers often avoid disclosing future designs. However, continuing silence would preserve the central uncertainty around Intel’s commercial foundry ambitions.

TSMC’s progress provides the competitive reference for all three signals. Its 2028 production plan is supported by established customer relationships and a long record of manufacturing for outside designers. Any Intel delay gives those customers another reason to stay with their existing supplier.

Intel does not need to match every part of TSMC’s business by 2028. It needs to show that 14A offers enough performance, capacity, service, and predictability to win meaningful designs. That standard is demanding but measurable.

For PC buyers, the near-term issue is whether tight supply and higher average prices persist. For data-center operators, Intel’s product cadence and factory recovery can influence server availability, competition, and infrastructure planning.

Chip designers face a larger strategic choice. A credible Intel Foundry would add advanced manufacturing capacity in the United States and reduce dependence on one supplier. An unreliable alternative would add engineering cost without delivering useful diversification.

The next few quarters will reveal which interpretation deserves more weight. Watch whether margins improve with revenue, whether 18A scales cleanly, and whether an external customer commits to Intel 14A. Those signals will show whether the Tom Hardware turnaround story is becoming a manufacturing reality or remaining an ambitious timetable.