Basis Report/Resources/Investor Foundations
4 sections20 entries

Semiconductor Earnings Questions That Reveal the Cycle Turn

Every semiconductor earnings call has a story and a different story underneath it. The story is revenue growth, gross margin recovery, and improving book-to-bill. The story underneath is whether the demand is real or channel-driven, whether the utilization recovery is structural or cyclical, and whether the next quarter's guide is anchored in actual customer commitments or in management's confidence about the recovery timeline.

✓Before the call, write down your estimate of book-to-bill and channel inventory levels by end market — then check the disclosed data specifically and ask why they diverge from your estimate.
✓Strip utilization-driven gross margin recovery from structural mix and pricing improvement before deciding the margin trajectory deserves a premium multiple.
✓Ask what percentage of the sequential revenue guide is covered by take-or-pay commitments versus spot purchasing before modeling the guide as a floor.
✓Track foundry utilization disclosures from TSMC and Samsung alongside the company's own guidance — divergence between the two is usually the first signal that the channel inventory situation is different than management is describing.
Go to frameworkSee a sample report
When to use this

Use this before every semiconductor earnings call, when reviewing the transcript two days after the call, and when the market reacts to results in a way that seems disconnected from the underlying demand signals. The most important preparation is the pre-call analysis of foundry utilization, book-to-bill trends, and channel inventory levels — not the post-reaction positioning.

Why it matters now

AI infrastructure investment has complicated the traditional semiconductor cycle framework. Hyperscaler demand for AI accelerators has created a parallel cycle that is partially insulated from the consumer electronics and PC cycles but is not immune to its own inventory and capacity dynamics. Understanding which demand signals apply to the AI-exposed portion of a semiconductor company versus the traditionally cyclical portion is the central analytical challenge right now.

Where theses break

The question bank only works if you are willing to update the cycle thesis when the leading indicators are pointing in a different direction than management's commentary. The most common failure mode is confirming a recovery thesis with the improving headline numbers while ignoring the channel inventory signals that are telling a different story two quarters out.

Full framework

4 sections · 20 entries — work through each before you size a position.

The semiconductor industry is one of the most systematically over-optimistic in earnings guidance. At cycle tops, management teams describe strength that leads to disappointment. At cycle troughs, they describe weakness that leads to upside. The most useful work is not interpreting the current quarter's beat or miss — it is understanding where the cycle is and building a framework for what the next six quarters will look like before consensus does.

20 entries in view

Questions that reveal whether demand is real or channel-driven

Semiconductor revenue can look strong while real end demand stays weak — these questions expose the difference between sell-in and sell-through before you build any model on the headline number.

Establish whether the revenue beat came from end demand or distributor restocking before modeling any forward acceleration

Ask explicitly whether the sequential revenue improvement came from stronger sell-through — actual end customers consuming more — or from distributors refilling depleted channel inventory after a period of inventory correction. A sell-in beat that is not confirmed by sell-through data means the inventory correction may be extending forward rather than ending. Distributors refilling their shelves is not the same as OEMs increasing production plans, which is not the same as end consumers purchasing more. The demand chain is three to four steps long, and beats at the first step — manufacturer to distributor — do not automatically indicate recovery at the last step.

Why it matters

The semiconductor revenue beat is the single most misleading data point in the industry because it reflects sell-in, not sell-through. A company can beat revenue estimates while real demand is still deteriorating — channel inventory is simply being rebuilt at the distributor level rather than destroyed at the end customer level.

When it matters

Immediately after reading the revenue line in the press release, and specifically in any quarter following a period of inventory correction where distributors had been working down excess inventory.

Investor take

Ask management directly: 'Can you confirm whether the revenue upside reflected stronger sell-through at the end customer level, or primarily sell-in to replenish distributor inventory?' Any answer that does not distinguish between the two is not sufficient to model a recovery thesis.

Check the book-to-bill ratio and its trend before crediting a demand inflection

The book-to-bill ratio measures the dollar value of orders received versus the dollar value of product shipped in a given period. A ratio above 1.0 means more was ordered than shipped — a demand signal. But the ratio has been systematically distorted by supply chain behavior: during shortages, customers double-order to secure supply. During corrections, orders collapse while shipments continue from the backlog, pushing the ratio below 1.0 further than real demand would justify. The most useful reading is the trend over three or more quarters rather than any single quarter's level.

Why it matters

Book-to-bill is the canonical semiconductor demand indicator, but its signal quality has declined because supply chain disruptions have introduced ordering behavior that diverges significantly from real demand. A rising book-to-bill is a necessary but not sufficient condition for concluding that real demand is recovering.

When it matters

Every semiconductor earnings call. Most important in the quarter after a significant inventory correction or after a period of extended lead times, when the ordering behavior may be reverting from one extreme toward another.

Investor take

Track whether the trend in book-to-bill is consistent with distributor inventory levels reported in the same quarter. A rising book-to-bill alongside rising distributor inventories suggests that customers are ordering ahead of confirmed consumption. A rising book-to-bill alongside falling distributor inventories is a cleaner demand signal.

Ask what customer inventory weeks look like in each end market before accepting that the correction is over

Customer inventory weeks — the number of weeks of supply customers hold on their own balance sheets — is the leading indicator of whether OEMs are about to start ordering again or whether the correction has further to run. When customers have excess inventory, they stop placing new orders regardless of their own end demand, and semiconductor revenue falls. The correction ends not when end demand recovers, but when customer inventory falls to normalized levels. Ask for the inventory level in each end market specifically — the PC correction can be over while the industrial correction is still in progress, and treating them as a single inventory cycle will produce an incorrect forward model.

Why it matters

Customer inventory weeks is the specific metric that determines when order rates resume. It is often disclosed indirectly in the earnings call — customers describing inventory as 'near normal' or 'largely worked off' are providing the signal even without a specific number. Tracking this commentary across earnings calls is more reliable than waiting for the revenue recovery.

When it matters

During any period of inventory correction, and in the first two to four quarters after a demand downturn when the timing of the recovery is the most important modeling variable. Most important for cyclical end markets — PC, mobile, industrial, automotive — where inventory management behavior is more variable.

Investor take

Map the inventory situation in each of the company's end markets explicitly. Assign a rough status — excess, normalizing, normal, lean — to each. A company where all end markets are in the 'normalizing' stage is six to nine months from a revenue recovery. A company where two markets are still 'excess' is further out than any single metric will show.

Monitor lead time trends as the earliest available demand signal before orders convert to revenue

Lead times — the time between when an order is placed and when the product is confirmed for delivery — compress when supply is catching up with demand and extend when demand is outrunning supply. Because lead times are a real-time operational signal rather than a post-period accounting disclosure, they tend to lead the revenue inflection by one to two quarters. When a company that had been reporting twelve to eighteen week lead times begins describing lead times as 'normalizing toward historical levels,' the supply-demand balance is changing in a direction that will show up in revenue bookings before it shows up in recognized revenue.

Why it matters

Lead times are observable data points that precede revenue changes in both directions. Extending lead times are the earliest visible signal of a cycle turning up; compressing lead times are the earliest visible signal of a cycle turning down. Because they reflect current operational dynamics rather than backward-looking financials, they are the most actionable forward indicator in semiconductor earnings analysis.

When it matters

Every quarter, in both directions. Particularly important at cycle inflection points — when the market is uncertain whether a correction is over or a recovery is beginning. Lead time trends answer that question earlier than any financial metric.

Investor take

Read competitor earnings calls and foundry commentary alongside the company's own lead time disclosures. Divergences between what a chip designer says about demand and what its foundry says about utilization are early indicators that the demand picture the designer is presenting does not match the production reality.

Evaluate design win disclosures by end market to assess revenue visibility beyond the current quarter

Design wins — engineering decisions by OEMs to design a specific semiconductor into their next-generation product — convert into revenue only after the OEM's product enters mass production, typically twelve to twenty-four months later. A strong design win disclosure in a weak revenue environment provides visibility into future revenue that the current quarter's results do not reflect. A weak design win environment in a strong revenue quarter suggests that the current revenue base is less durable than it appears because the pipeline that feeds the next product generation is thinner.

Why it matters

Design wins are the deferred revenue of the semiconductor industry. They represent committed demand at the OEM product level, which is more durable than distribution channel demand because it reflects engineering decisions that are costly to reverse. A company accumulating design wins in automotive, industrial, and data center simultaneously is building a revenue pipeline that will sustain growth for three to five years regardless of the current quarter's cycle position.

When it matters

Every earnings call, but especially during cyclical downturns when the current revenue is depressed and the forward thesis depends on the strength of the design win pipeline rather than current orders.

Investor take

Build a simple design win pipeline tracker by end market: note the management commentary each quarter, whether the win rate is accelerating or decelerating, and which end markets are contributing. A company whose design win disclosures are getting more specific over time — naming customers, products, or platforms — is demonstrating stronger pipeline confidence than one whose disclosures are generic.

Numbers to reconcile before the quarter is fully understood

The earnings press release is the opening argument for semiconductors. These are the numbers that determine whether the argument holds when the utilization, inventory, and capex data are examined more carefully.

Reconcile the gross margin change against the utilization rate before crediting structural improvement

Semiconductor gross margins move primarily on two drivers: utilization rate and product mix. Utilization-driven margin improvement is cyclical — when a fab runs from 65% to 80% capacity utilization, the fixed cost per unit falls and margin expands without any change in pricing power, customer mix, or product capability. Mix-driven margin improvement reflects a shift toward higher-value products and is more durable. Before crediting gross margin expansion as structural, establish what percentage of the improvement came from utilization recovery versus mix and pricing. The utilization portion will compress again at the next demand deceleration.

Why it matters

Utilization-driven gross margin expansion is the most commonly misattributed source of semiconductor profitability improvement. At cycle troughs, every basis point of gross margin expansion gets credited to operational efficiency when it is actually a mechanical consequence of running the factory harder. Buying a semiconductor stock at cycle-peak utilization margins as if they were structural will produce significant downside surprise when utilization normalizes.

When it matters

Every time gross margin expands meaningfully year-over-year or quarter-over-quarter. Particularly important during the early and middle phases of an upcycle, when utilization is recovering from trough levels and the margin benefit is real but temporary.

Investor take

Build a simple sensitivity model: estimate what gross margin looks like at normalized historical utilization versus the current rate. If the company is running at 80% utilization and historical norms are 70%, the gross margin has five to eight percentage points of utilization-driven support that will compress when the cycle moderates. Model that scenario explicitly before valuing the business on peak margins.

Compare revenue by end market to the shipment mix before trusting the margin trajectory

Gross margin in semiconductors varies substantially by end market — data center and automotive parts generally carry higher margins than PC and mobile components because of lower volume, higher complexity, and more favorable competitive dynamics. When data center or automotive revenue rises as a percentage of the mix in a given quarter, gross margin can improve purely because of mix without any improvement in the underlying pricing or production economics of any individual product. Before concluding that the gross margin trajectory is strengthening across the board, verify that the mix shift is durable rather than a function of a specific quarter's product shipment timing.

Why it matters

End market mix shifts are a legitimate source of gross margin improvement, but they are also one of the most frequently cited explanations for margin beats that do not persist. A company where automotive revenue beat because of a single large shipment to one OEM will not sustain that margin in the following quarter unless the shipment pattern repeats.

When it matters

Any quarter where gross margin expanded and the commentary mentions favorable mix as a contributor. Also important when the company has recently shifted strategic emphasis toward a higher-margin end market where the early wins may be lumpy.

Investor take

Build a gross margin bridge: estimate the margin for each end market separately at normalized volumes, then calculate what the blended margin would be at normalized mix. Compare that to the actual reported margin. If the reported margin is materially higher than the normalized-mix estimate, the improvement is mix or timing-dependent and may not be repeatable at the same rate.

Check the inventory build in dollar terms versus units shipped before accepting that the balance sheet is clean

A rising inventory balance in dollars can indicate either that the company is building safety stock in anticipation of demand recovery, or that units are accumulating because demand fell below production levels and the company has not yet cut wafer starts to match. The distinction matters because inventory built in anticipation of recovery will convert to revenue if the recovery arrives on schedule, while inventory accumulated because demand fell short of plan is a risk that will either require a write-down or compress margins when excess units are sold at reduced prices.

Why it matters

Semiconductor companies have a history of characterizing rising inventory as strategic when it is actually the product of production commitments that outran order rates. The distinction between inventory-as-strategy and inventory-as-problem typically becomes clear two to three quarters later when the company either ships the inventory into an improving demand environment or writes down the portion that does not convert.

When it matters

Any quarter where the inventory balance rose faster than revenue, and in the quarters immediately following a demand deceleration when production cuts have not yet been fully implemented.

Investor take

Calculate inventory days on hand — inventory balance divided by the quarterly COGS divided by 90 days — and compare it to the five-year historical average. If the current level exceeds the historical average by more than twenty days, the inventory position requires a specific explanation of what demand is being reserved for and when it will ship.

Verify capex spend relative to when new capacity will actually be available before trusting the supply projection

Semiconductor capital expenditures — particularly fab construction and advanced equipment purchases — have extremely long lead times, often eighteen to thirty-six months from commitment to revenue-generating capacity. When management announces an increase in capex, the capacity enabled by that investment will not be available for production for at least one to two years. In an environment where management is simultaneously describing near-term supply constraints, the capex ramp will not relieve those constraints within the current fiscal year.

Why it matters

Capex guidance is consistently misread as near-term supply relief when it is actually a multi-year investment with deferred revenue implications. An investor modeling that a $2B capex program announced today will relieve supply constraints in the next quarter is making a timing error that will not be visible in the numbers for twelve to eighteen months.

When it matters

Every time management announces a significant capex increase alongside a discussion of supply constraints or demand recovery. Also important when evaluating whether a foundry's capacity build will relieve customer lead times in the near term versus the medium term.

Investor take

Build a capacity ramp timeline: for every major capex commitment in the last three years, note the estimated online date and production ramp schedule. Use this to estimate when incremental capacity will actually affect quarterly production and revenue. This is different from when the cash is spent, and the difference can be significant.

Assess the non-GAAP to GAAP gross margin spread before accepting the profitability narrative

Semiconductor companies frequently report non-GAAP gross margins that exclude stock-based compensation, amortization of acquired intangibles, restructuring charges, and facility-related one-time costs. The spread between non-GAAP and GAAP gross margin can be six to twelve percentage points for a company that has completed a significant acquisition or that runs a large equity compensation program. Before building a multiple on the reported non-GAAP gross margin, verify that the excluded items represent genuinely non-recurring costs rather than recurring costs that management prefers to report separately.

Why it matters

The non-GAAP gross margin is the metric management promotes and the metric most analysts use for peer benchmarking. But if the company consistently excludes three to five percentage points of real costs — equity grants that dilute shareholders, acquisition integration costs that recur every few years — the margin the business actually generates for owners is materially below the reported figure.

When it matters

Whenever the reported non-GAAP gross margin is being used to justify a valuation premium or a peer comparison. Also when evaluating the sustainability of the margin at a level that would require the GAAP adjustments to remain constant while revenue scales.

Investor take

Build your own adjusted gross margin line: take the GAAP gross profit and add back only the amortization of acquired intangibles. Do not add back SBC, restructuring, or facility charges unless they were explicitly disclosed as genuinely one-time. Compare that number to both the non-GAAP figure management promotes and to the GAAP figure — it is the most honest representation of the business's profitability at the gross margin level.

Forward signals that determine the next leg of the cycle

Semiconductor cycles move in waves that are visible in the data before they appear in the revenue line. These are the signals worth tracking between earnings calls.

Assess the end market composition of the backlog before extrapolating current growth rates forward

The semiconductor backlog is only as durable as the end market demand that created it. A backlog dominated by automotive or industrial orders tends to be stickier because procurement decisions in those industries involve longer design cycles and supply chain commitments that are costlier to cancel. A backlog dominated by consumer electronics or PC orders is more likely to be revised when end demand shifts, because OEM inventory management in those end markets is more aggressive. Before extrapolating current revenue growth rates into the next two to four quarters, examine the backlog composition and ask whether the end markets driving it have historically honored their order books through demand shifts.

Why it matters

Semiconductor revenue guidance is based on the backlog and the conversion of orders into shipments. If the backlog is vulnerable to cancellation or order pushout in one or more end markets, the guidance accuracy will reflect that vulnerability before the revenue line confirms it.

When it matters

Every quarter, and especially during the first signs of macro weakness when the risk of OEM inventory management behavior increases. Also important when the company has recently shifted its end market exposure through acquisitions, new product lines, or capacity allocation decisions.

Investor take

Ask management to characterize the backlog by end market and by the mix of take-or-pay commitments versus cancellable orders. A company with a 70% take-or-pay backlog has a much higher floor on the next quarter's revenue than one with 30% take-or-pay coverage.

Evaluate AI and data center capacity commitments versus actual measured demand pull before calling the upcycle durable

Hyperscaler capital expenditure on AI infrastructure represents capacity investment that is made ahead of confirmed application utilization. Hyperscalers are building compute capacity in advance of the workloads that will fill it because they expect those workloads to grow and because the lead times for acquiring compute are long. This creates a difference between demand in the hyperscaler capital deployment sense and demand in the end user application utilization sense. When application utilization catches up with deployed capacity, the pace of incremental capacity deployment may slow even if AI workload growth continues.

Why it matters

The distinction between capital deployment demand and utilization-driven demand determines the durability of the current AI semiconductor cycle. If hyperscalers are buying capacity in advance of measured utilization, there will be a period of absorption when new ordering moderates while utilization catches up. That absorption period will look like a demand deceleration even if AI application growth continues at a high rate.

When it matters

Every quarter in the AI-exposed portion of a semiconductor company's business, and specifically when hyperscaler capex guidance shows any sign of moderation. Also important when a semiconductor company's AI-driven revenue growth rate decelerates.

Investor take

Track measured AI compute utilization separately from hyperscaler capex guidance. Use earnings calls from cloud providers to estimate the ratio of AI compute capacity to AI compute utilization. When that ratio is expanding — more capacity being built than is being utilized — the risk of a digestion period in the next two to four quarters increases.

Monitor wafer start disclosures and foundry partner commentary as a leading indicator of the company's own production plans

Wafer starts — the number of silicon wafers put into production at the foundry — are disclosed by TSMC and other foundries in their earnings calls and financial disclosures, and represent the leading indicator of the chip supply that will be available for shipment in three to four months. A semiconductor company's own guidance does not exist in isolation from the foundry utilization data — if the foundry is describing weakness in advanced node capacity utilization while the chip company is guiding for sequential growth, the divergence requires explanation.

Why it matters

Foundry utilization data is among the most externally verifiable signals in semiconductor earnings analysis because it comes from a different company with independent interests in providing accurate operational disclosure. When a semiconductor company's revenue guide implies production volumes that are inconsistent with what the foundry is disclosing about utilization, one of the two is incorrect.

When it matters

Every quarter, after TSMC's monthly revenue data and quarterly earnings call. Particularly important when a fabless semiconductor company is providing guidance that relies on increasing wafer allocations from its foundry partners.

Investor take

Read the TSMC and Samsung earnings calls alongside any semiconductor company you are evaluating. Build a simple utilization comparison: what is the foundry disclosing about capacity utilization, and what does the chip company's guidance imply about wafer volumes? Divergence between the two deserves an explicit reconciliation rather than passive acceptance.

Track automotive and industrial semiconductor content per unit as the durable secular signal beneath the cyclical noise

The long-term secular growth thesis in semiconductors is built on content growth — the rising dollar value of semiconductors per end unit. A modern electric vehicle contains $600 to $1,000 of semiconductor content per vehicle, versus $200 to $400 for a traditional internal combustion vehicle. Industrial automation equipment, power infrastructure, and factory robotics are all increasing their semiconductor content per unit as systems become more intelligent and connected. Even in a year when automotive production volumes are flat, semiconductor content per vehicle can grow 8% to 12%.

Why it matters

The semiconductor cycle is mean-reverting — unit volumes in any end market will eventually recover from any given demand trough. But the content growth opportunity is one-way — as vehicles and industrial equipment get more complex, the semiconductor content per unit grows. An investor who confuses a cyclical automotive production trough with a secular decline in semiconductor demand is making the same mistake as one who confuses a cyclical recovery with a permanent expansion.

When it matters

Every quarter, as a framework for evaluating management commentary about automotive or industrial market outlook. Also important when valuing a semiconductor company whose customer exposure is concentrated in one or both of these end markets.

Investor take

Build a simple content-per-unit model: estimate the company's semiconductor dollar value per vehicle or per industrial unit in the current year and in the prior two years. Compare the growth rate in content-per-unit to the growth rate in total automotive or industrial revenue. If content-per-unit is growing while total revenue is flat or declining, the secular opportunity is real but is being masked by cyclical volume weakness.

Assess node transition timing against competitor capacity to determine whether the technology leadership advantage is widening or narrowing

The most durable competitive advantage in semiconductor design is being two to three generations ahead of competitors at the process node that matters for the target end market. Node transitions — moving from 5nm to 3nm or from 3nm to 2nm — are expensive, complex, and subject to delays, but the first company to achieve high-yield production at an advanced node typically enjoys eighteen to twenty-four months of gross margin advantage before competitors close the gap. Ask where the company is in its node transition roadmap, where competitors are, and what the yield and cost picture looks like at the current leading node.

Why it matters

Node leadership is the highest-quality source of semiconductor competitive advantage because it translates directly into product performance at equivalent cost, which determines design win outcomes at OEMs who need the best performance-per-watt available. A company that loses node leadership typically loses design wins in the next generation cycle, which is visible in the design win disclosures before it shows up in revenue.

When it matters

Any time a competitor announces a significant node advancement, when a foundry announces a new node is entering mass production, and whenever management discusses R&D spending and product roadmap in the context of the competitive environment.

Investor take

Track the node generations of all major competitors alongside the company's own roadmap. When evaluating a new product announcement, ask what node it uses, what node the primary competitor's equivalent product uses, and what the performance and cost difference implies for design win probability. A company whose new product uses a node one generation behind the competitor's latest offering is in a weaker competitive position than the announcement alone implies.

How to frame the next quarter's debate before consensus hardens

The three months between semiconductor earnings are where the cycle thesis is tested. These questions set the agenda for what to track and what to decide before the next results arrive.

Determine whether the guidance raise was driven by pricing power, volume improvement, or mix shift before modeling it forward

When a semiconductor company raises forward revenue guidance, the source of the increase matters more than the magnitude. A guidance raise driven by pricing improvement — higher ASPs because supply is constrained relative to demand — is cyclical and will compress when supply capacity catches up. A guidance raise driven by unit volume improvement is a stronger signal because it reflects the demand environment rather than supply dynamics. A guidance raise driven by mix shift toward higher-value end markets is meaningful but requires verification that the mix shift is structural rather than a temporary concentration in a high-margin segment.

Why it matters

Semiconductor ASP cycles are highly predictable once supply-demand is understood. Pricing tends to peak when lead times are longest and capacity is tightest, and to fall rapidly once wafer supply increases and lead times normalize. A guidance raise built entirely on ASP improvement will produce a guidance cut twelve to eighteen months later when supply responses close the supply gap.

When it matters

Every time guidance is raised, but especially in late-cycle quarters when the beat and raise are occurring while lead times are still extended and distributor inventory is still below normal.

Investor take

Build a guidance attribution model: estimate what percent of the guidance raise came from ASP versus units versus mix. Use the foundry commentary on capacity additions to estimate when supply relief is likely, and ask what ASP the guidance implies at that supply level. A guidance raise built 70% on ASP improvement in a constrained supply environment has a different two-year outlook than one built 70% on unit volume improvement.

Evaluate where gross margin sits relative to normalized utilization before deciding whether margin expansion is priced in

Semiconductor gross margin is highly cyclical, and the current margin level relative to normalized utilization tells you how much margin support is already embedded in the current stock price. If a company is reporting 58% gross margin while running at 85% utilization, and historical normalized utilization is 70-75%, the gross margin will compress toward 50-53% when utilization returns to normal levels — not because anything deteriorated, but because the utilization-driven cost absorption will reverse.

Why it matters

The semiconductor multiple is most expensive at the peak of the cycle precisely when it appears most justified by the current margin and revenue data. A company at 58% gross margin and 20x forward earnings looks cheap if you assume the margin is structural. The same company at 50% gross margin — which is what normalized utilization would produce — trades at 28x on the same earnings base.

When it matters

Any time a semiconductor stock has re-rated higher based on margin expansion that occurred during a period of capacity tightness, and specifically in the late phases of an upcycle when utilization rates are at their highest and the cycle turn is closest.

Investor take

Build a normalized margin model: using the company's most recent historical trough margin as the floor and the current peak as the ceiling, estimate what margin looks like at the midpoint of the utilization range. Value the stock on the normalized margin rather than the peak margin, and check whether the current price implies the peak margin is permanent before taking a position.

Assess the long-term supply agreement coverage versus spot exposure before assuming pricing is locked in

Some semiconductor companies have secured multi-year supply agreements with pricing provisions that provide revenue and margin stability above the spot market. These agreements — common in automotive, industrial, and data center end markets — mean that a portion of the revenue base is insulated from near-term pricing pressure even when supply conditions shift. Other companies sell predominantly through distribution channels at spot pricing, which means any supply-demand normalization immediately flows through to realized ASPs.

Why it matters

The presence or absence of long-term supply agreements fundamentally changes the risk profile of a semiconductor company at any given point in the cycle. A company with 60% of revenue covered by multi-year pricing agreements has meaningful downside protection when spot pricing falls. A company selling entirely through distribution at spot pricing will see ASPs compress within one to two quarters of any supply normalization.

When it matters

When evaluating guidance credibility in the context of a cycle turn, and when a semiconductor company's reported ASPs are materially above the spot market for equivalent products — a condition that implies either pricing agreement protection or a product mix advantage that may not be visible in the aggregate numbers.

Investor take

Ask management what percentage of forward revenue is covered by pricing agreements and what the average duration is. Compare the implied ASP in the guidance to spot market references for similar products. A material premium to spot pricing implies either long-term agreement protection that should be disclosed or a product mix advantage that should be explained.

Write down what a demand shortfall would look like in the next quarter and which metric would signal it first

The most useful exercise after a semiconductor earnings call is defining in advance what a demand shortfall would look like in the following quarter and which leading metric would signal it first. Typically, the sequence is: lead times compress, book-to-bill falls toward or below 1.0, distributor inventory begins building, guidance is revised lower, revenue misses. By the time revenue misses, the earlier warning signals have typically been visible for one to two quarters.

Why it matters

Semiconductor cycle management depends on identifying the inflection points before they are confirmed by the lagging revenue data. Most investors respond to revenue misses rather than to the leading indicators that preceded them. Defining the warning signals in advance creates a decision framework that allows earlier action, either reducing risk before the revenue confirms the correction or increasing exposure when the leading indicators signal recovery ahead of the revenue inflection.

When it matters

Immediately after every earnings call, before the next quarter's results arrive. Most important at cycle peaks — when the current-quarter data is most compelling and the risk of complacency is highest — because that is precisely when the leading indicators are beginning to diverge from the revenue data.

Investor take

Write the following after each semiconductor earnings call: 'The next quarter is showing early signs of demand weakness if book-to-bill falls below X, if lead times compress by more than Y weeks, or if distributor inventory builds by more than Z days relative to the current quarter.' Use those specific thresholds as an active monitoring framework rather than waiting for the revenue data to confirm what the leading indicators are already signaling.

Identify where analyst consensus is assuming a different utilization path than the company's own production signals imply

Semiconductor analyst consensus models are built largely on management guidance, which reflects the company's view of demand at the time of the call. But the consensus utilization assumption embedded in those models may diverge meaningfully from what the foundry utilization data, the capital equipment delivery schedules, and the distributor inventory reports are collectively implying. When consensus is modeling 80% utilization for the next four quarters and the foundry data implies utilization is already tracking toward 70%, the consensus gross margin estimate is too high and the consensus EPS estimate will need to be revised.

Why it matters

Analyst consensus is backward-looking in semiconductor cycles because it updates based on the most recent management guidance rather than on the leading indicators that precede the guidance revision. The most significant earnings surprises in semiconductor stocks — both positive and negative — occur when the leading indicators were pointing in a different direction than consensus for one to two quarters before the revision arrived.

When it matters

When the consensus forward model implies a utilization rate that is materially different from what the supply chain data suggests, and specifically when the consensus is extrapolating the current quarter's results into the next two to four quarters without a scenario for what a utilization compression would look like.

Investor take

Build a parallel model alongside consensus: one that uses the company's guidance and one that uses the foundry utilization data and lead time trends as inputs. When the two models converge, consensus is likely to be approximately right. When they diverge materially, there is a variant perception opportunity — either the consensus is too optimistic or too pessimistic, and the leading indicators tell you which.

Evidence

Semiconductor earnings scorecard

Six metrics to reconcile before the quarter is fully understood

The headline revenue and gross margin numbers are the opening argument. These six metrics are the cross-examination. When they conflict with the headline story, trust the conflict.

Book-to-Bill Ratio
Orders received ÷ revenue shipped in the same period
The most direct measure of whether backlog is building or drawing down. Above 1.0 means orders are coming in faster than the company can ship — demand is outpacing supply. Below 1.0 means the company is burning through backlog without replenishing it. A single quarter above 1.0 is not a cycle signal; two or three consecutive quarters above 1.0 accompanied by lead time expansion is the historical pattern that precedes meaningful revenue inflections.
Fab Utilization Rate
Actual production output ÷ total nameplate capacity
The single biggest driver of gross margin in companies that own their own fabs. At 90% utilization, fixed costs are spread across maximum units — margins expand. At 60% utilization, the same fixed costs are absorbed by fewer units — margins compress sharply. A gross margin improvement needs to be decomposed into utilization recovery versus structural product improvement; only the structural component deserves multiple expansion credit.
Channel Inventory (Weeks of Supply)
Distributor/OEM inventory ÷ weekly consumption rate
Tracks how many weeks of supply sit in the channel between the chipmaker and the end customer. Normal operating levels are typically four to eight weeks depending on the end market. When channel inventory exceeds twelve to sixteen weeks, customers stop ordering at the consumption rate — which creates revenue headwinds even when end demand is stable. This is the primary mechanism by which semiconductor inventory corrections unfold.
Lead Times by Product Family
Time between order placement and confirmed ship date
A leading indicator of supply-demand balance, typically one to two quarters ahead of revenue. Expanding lead times signal that orders are outrunning available production — an early upcycle signal. Compressing lead times signal that supply is catching up to or exceeding demand — the first warning of a potential correction. Track the direction rather than the absolute level, and compare across competitors in the same end market.
Inventory Days on Hand
Inventory balance ÷ (COGS ÷ 90 days)
Tracks how many days of production sits on the company's own balance sheet. Rising inventory days is a yellow flag — the company may be producing ahead of demand, accepting chips from its foundry at the contracted rate while its own customers are pulling back. When inventory days exceeds historical norms by more than thirty days and gross margin is also under pressure, the risk of an inventory write-down in subsequent quarters increases materially.
Revenue by End Market (Mix Trend)
Quarterly revenue split: data center, auto, industrial, mobile, PC, other
Each end market has a different margin profile, growth rate, and cycle duration. Data center and automotive typically carry higher gross margins than PC and mobile. A gross margin beat that comes from favorable end market mix may not be repeatable if the mix was driven by timing or temporary concentration. Track the mix over eight quarters rather than focusing on the current quarter's composition, and flag when any single end market represents more than 35% of revenue.

Demand signal diagnostic

What the headline beat actually tells you — and what you need to check underneath

Semiconductor beats come from several different sources with different implications for the next two quarters. Identifying the source is more important than the magnitude.

What the headline beat actually tells you — and what you need to check underneath
Demand signalSurface readWhat it might actually beWhat to verify
Revenue beat on strong sequential growthDemand is recovering and the cycle is inflecting upwardChannel restocking after an inventory correction — distributors refilling shelves, not end customers buying moreSell-through data from distributors; end market unit production volumes from automotive or PC data; whether the company's own channel inventory commentary changed this quarter
Book-to-bill above 1.0 for the first time in several quartersThe upcycle is beginning — a reliable historical buy signalSafety stock ordering — customers booking aggressively to protect supply after a period of shortages, which creates double-ordering that is not all real demandTwo consecutive quarters of book-to-bill above 1.0; lead times expanding alongside the book-to-bill rather than staying flat; whether the order composition has changed materially toward longer-duration commitments
Gross margin expansion of 200+ basis points quarter-over-quarterStructural margin improvement — the business quality is strengtheningUtilization recovery — the same fixed cost base is being spread over more units because the fab is running harder, not because the product mix has improved or pricing has strengthenedDisclosed utilization rate versus prior quarter; whether the company's management specifically attributes the margin improvement to utilization versus mix or pricing; whether a fab underloading charge is no longer recurring
Data center revenue growing 40%+ year-over-yearAI infrastructure demand is a multi-year secular growth driver that will sustain elevated capital spendingConcentrated hyperscaler purchasing that reflects capital deployment rather than application demand — hyperscalers are buying capacity ahead of confirmed utilization, creating a period of absorption when ordering moderatesRevenue distribution across the top three or four hyperscalers; whether customer concentration is rising or falling; management commentary on whether orders reflect current utilization needs or forward capacity planning
Inventory write-down taken in the current quarterManagement has cleaned up the balance sheet — a one-time charge that signals the inventory problem is resolvedA lagging indicator that the inventory problem is more persistent than one quarter of write-downs implies — write-downs usually follow demand weakness by two to three quarters and often occur in stagesThe write-down amount as a percentage of the total inventory balance; whether days of inventory outstanding fell meaningfully or only modestly; management guidance on when they expect inventory to normalize to historical levels

Common questions

What investors ask about investor foundations for investor foundations stocks.

How do I know if a semiconductor revenue beat reflects real end demand or channel restocking?
The most reliable test is comparing sell-in — what the chipmaker shipped — to sell-through — what end customers actually consumed. When sell-in is growing faster than sell-through, the difference is inventory build at the distributor or OEM level rather than consumption growth. If the company does not disclose sell-through data directly, use the distributor earnings calls and inventory level disclosures from customers to triangulate. A revenue beat that is not confirmed by sell-through data is a channel-fill beat — useful for the current quarter but not a signal of sustained demand recovery.
What is the most reliable early warning signal that a semiconductor upcycle is turning?
Lead time compression is typically the earliest visible indicator — it shows up one to two quarters before book-to-bill falls below 1.0, and two to four quarters before revenue guides start moving down. When a company that has been reporting extended lead times starts describing lead times as 'normalizing,' that is the first signal worth monitoring closely. If lead time compression is accompanied by rising distributor inventory levels and the company's own inventory build, the probability of a demand correction in the next two to three quarters increases materially. These three indicators together — lead times compressing, distributor inventory rising, company inventory building — form the standard early warning pattern.
Why do semiconductor gross margins move so much between cycles?
Gross margins in semiconductor manufacturing are highly sensitive to fab utilization because the cost structure is predominantly fixed — depreciation, labor, and facility costs do not change whether the fab is running at 60% or 90% capacity. At 90% utilization, those fixed costs are spread across 50% more units than at 60%, which dramatically reduces cost per unit and expands gross margin. The range between trough-utilization margins and peak-utilization margins for an IDM — integrated device manufacturer — that owns its own fabs can exceed fifteen to twenty percentage points. For fabless companies using contract manufacturing, the sensitivity is lower but still meaningful because wafer pricing from foundries also fluctuates with supply-demand conditions.
How should I think about AI-driven semiconductor demand differently from traditional semiconductor cycles?
AI accelerator demand is driven primarily by hyperscaler capital expenditure decisions rather than by consumer end demand, which means the demand signal is a capital allocation decision by a small number of large customers rather than an aggregate consumption signal from millions of end users. This creates a different risk profile: demand can be more concentrated, more sensitive to hyperscaler spending cycle adjustments, and more subject to temporary pauses when deployed capacity temporarily outpaces measured application utilization. The secular growth in AI infrastructure is real and likely to be very large, but the path is not linear — it goes through periods of digestion and inventory normalization just as traditional semiconductor cycles do, with the additional complexity that the digestion period is harder to predict because the demand signal is internal to the hyperscaler rather than visible in consumer sales data.

Continue research

Keep building the foundation

Semiconductor stock analysis: reading the cycle before repricingInvestor FoundationsSemiconductor Stock Valuation: Underwriting the CycleInvestor FoundationsEarnings quality: score the cash before you trust the beatInvestor FoundationsGAAP vs adjusted earnings: audit every add-backInvestor FoundationsEarnings report autopsy: what the press release buriesInvestor FoundationsTech multiples: what actually deserves a premiumInvestor Foundations

Put it on a live stock

NVDA stock overviewJump to the live ticker page and report flowAMD stock overviewJump to the live ticker page and report flowAVGO stock overviewJump to the live ticker page and report flowAMAT stock overviewJump to the live ticker page and report flow

Category hub

Earnings analysis hubSee all guides in this category