Nvidia vs Micron stock: cash flow tests the cheaper AI bet

Nvidia vs Micron stock is no longer a choice between a fresh buyback announcement and an earnings report still to come. Micron has delivered its September 30 results. The more useful question now is how much of each company's extraordinary profit becomes cash that shareholders can actually keep. On a consistent cash-after-capital-purchases calculation, Micron offers roughly 5.8% of its reconstructed equity value, against 2.4% for Nvidia. That is a meaningful valuation advantage, but it needs a test of durability.

The filings reveal different pressures behind those numbers. Nvidia has extended payment terms on some large customer agreements, tying up money in receivables. Micron has obtained substantial customer deposits, but most of those deposits are expected to be returned later and are recorded in financing cash flow. Neither a sale awaiting collection nor a refundable deposit is the same thing as unrestricted profit. That distinction changes how investors should read the familiar argument that Nvidia has the stronger business while Micron has the cheaper stock.

Nvidia vs Micron: the decision in brief

Micron has the stronger trailing valuation case; Nvidia has the stronger platform argument. For an investor prepared to monitor memory pricing, manufacturing investment and contract delivery, Micron's discount is large enough to deserve serious consideration. For someone prioritizing a broadly integrated computing platform, Nvidia's hardware, networking and software offer a different source of competitive strength. Its premium still requires better future cash generation, not just a large repurchase authorization.

This is a comparison of two businesses available in TECHi's stock directory, not a prediction that one must outperform next week. Buying both also does not remove their common exposure to AI infrastructure spending. A customer spending slowdown, a shortage of available power or a financing problem in the data-center chain can reach both companies through different channels.

Start with matching valuation definitions

At publication, the fixed price observations are the October 19, 2026 regular-session closes: $217.09 for NVDA and $904.90 for MU, in U.S. dollars. TECHi's Nvidia quote page and Micron quote page show Yahoo Finance observations at 4:00 p.m. EDT; the underlying timestamps differ by one second. The pages label prices delayed, without specifying a fixed delay. These are closing snapshots, not live after-hours prices.

For a reproducible equity-value denominator, multiply those prices by each company's last reported period-end common shares outstanding: 24.147 billion for Nvidia and 1.131 billion for Micron. That produces approximately $5.242 trillion and $1.023 trillion. These are TECHi reconstructions using dated share counts, not claims that every share issued or repurchased since those filings has been captured. They can differ from a quote provider's current market capitalization.

The earnings and cash-flow comparison uses each company's latest four reported fiscal quarters. Nvidia's window ends July 26, 2026; Micron's ends September 3, 2026. Both are annual reporting windows, but they are not identical calendar periods, and Micron's fiscal 2026 contained 53 weeks. Exact calendar alignment is not available from these quarterly disclosures without making estimates. We therefore identify the mismatch instead of comparing a full year for one company with a single quarter for the other.

Nvidia's trailing totals are reconstructed from its fiscal 2026 release, adding the first half of fiscal 2027 and subtracting the first half of fiscal 2026. Revenue is $215.938 billion plus $177.837 billion minus $90.805 billion, or $302.970 billion. The same method gives $192.880 billion of GAAP net income and $134.360 billion of operating cash flow. Micron's latest annual filing supplies its four-quarter totals directly.

For cash after capital purchases, the same rule applies to both businesses: operating cash flow minus purchases of property, plant and equipment, including intangible purchases where Nvidia combines them with equipment. This produces $127.006 billion for Nvidia and $58.963 billion for Micron. It excludes acquisitions, securities purchases, debt repayments and dividends. It also excludes asset-sale and government-grant inflows. It is a comparison measure, not cash legally available for unlimited distribution or either company's identically defined reported free cash flow.

That choice matters. Micron's reported adjusted free cash flow includes offsets to capital spending, including government incentives. Nvidia's own free-cash-flow definition also subtracts specified principal payments. Using the companies' headline labels without checking their reconciliations would give an appearance of precision while mixing definitions. Our narrower common calculation makes the trade-off easier to inspect.

What an investor owns in each business

Nvidia sells accelerated computing as a system. Processors matter, but so do networking, software libraries and the work required to deploy and operate large clusters. Customers choosing a platform must consider engineering time and workload compatibility alongside a chip's purchase price. That can create switching costs and support repeat business; it does not eliminate competing accelerators or customers' efforts to develop their own silicon.

Micron manufactures memory and storage. DRAM holds data that processors need to access quickly; NAND provides persistent storage. High-bandwidth memory, or HBM, stacks memory close to accelerators to move large amounts of data efficiently. The economic attraction is that a more capable AI system can need more sophisticated memory as well as more compute. The economic risk is that memory manufacturers must commit substantial capital before all the future selling prices are known.

These companies are therefore not interchangeable competitors. Micron can benefit from demand across several accelerator ecosystems, while Nvidia benefits from the adoption of its overall computing platform. A comparison based only on which chip is more exciting misses the investment question: what return does the shareholder receive after production costs, reinvestment, financing obligations and the price paid for the stock?

Nvidia's premium meets a cash-collection test

Nvidia's latest quarter was exceptionally profitable: revenue reached $96.221 billion and GAAP net income $59.688 billion. Yet operating cash flow was $24.077 billion. Those figures from its August 26 earnings release do not establish that profits are unreliable. They establish that accounting earnings and cash arrived on different schedules.

Its July 26 Form 10-Q explains that extended payment terms on large, multi-quarter agreements with certain investment-grade customers contributed to the increase in receivables. The balance rose from $38.466 billion in January to $63.059 billion in July, roughly 64%. Investment-grade counterparties reduce some credit concerns, but extending terms still consumes working capital until collection occurs.

There is also an earnings-quality distinction beyond collection. The first-half cash-flow reconciliation removes $23.707 billion of gains from equity securities from net income. Such gains may increase the value of Nvidia's investments, but they are not the same as cash generated by selling computing systems. This helps explain why a GAAP earnings multiple alone can make the operating business look less expensive than a cash-based comparison suggests.

The strongest counterargument is timing. Nvidia made two federal income-tax payments in its second quarter and none in the first, according to the filing. A quarter with unusually heavy tax payments or working-capital investment need not represent the next quarter. That is why the main valuation uses four quarters. An improvement in collections, sustained customer demand and a stronger cash conversion rate would strengthen Nvidia's case even without a higher earnings multiple.

A practical test is to watch whether receivables growth moderates relative to revenue over subsequent reports. Investors should also check whether reported profit continues to rely materially on investment gains. A growing business can justify investment in customer relationships, but the justification becomes stronger when those relationships produce collections rather than repeatedly longer payment schedules.

Micron's contracts improve visibility, with important limits

Micron's September 30 earnings release reported $54.229 billion of fourth-quarter revenue and an 86.8% GAAP gross margin. Revenue exceeded its earlier $50 billion guidance midpoint by about 8.5%. Its next-quarter revenue outlook was $61.5 billion, plus or minus $1.5 billion, implying approximately 13.4% sequential growth at the midpoint. Those are management expectations for a future quarter, not reported sales.

The more consequential disclosure for a long-term comparison appears in Micron's fiscal 2026 Form 10-K. Strategic customer agreements include binding, multi-year volume commitments. Most have minimum and maximum pricing bands. This provides more contractual visibility than a thesis resting entirely on whatever spot memory prices happen to be next month.

However, the reported $134 billion of remaining performance obligations should not be treated as $134 billion of immediate revenue or cash. Roughly one-quarter is expected to become revenue over the next twelve months, with the balance expected within five years. The obligations use committed volumes and minimum pricing. Agreements without fixed prices or pricing bands are excluded because their consideration is variable. The number is a defined accounting disclosure, not a complete forecast of all future business.

Deposits require equally careful treatment. Micron recorded $12.747 billion of customer contract deposit proceeds in financing activities during fiscal 2026. Its accounting policy says deposits are returned when customers meet minimum purchase commitments; it may retain some or all when they do not. Nearly all deposits are scheduled for repayment between 2029 and 2031. They strengthen today's liquidity, but an investor cannot count them as both permanent shareholder cash and a promise of future sales.

Crucially, those financing inflows are already outside operating cash flow. Our $58.963 billion cash-after-capital-purchases calculation therefore does not include the deposit proceeds. Subtracting them again from that calculation would be another accounting mistake. Their relevance is to liquidity, future obligations and the interpretation of contract strength, not to an invented reduction in this year's operating cash flow.

The bear case remains a decline in memory profitability as supply catches up, customers change procurement plans or the mix shifts. The countercase is stronger than simply saying that this cycle is different: enforceable volume commitments and pricing floors could make a downturn less severe. What remains uncertain is the eventual cash margin after manufacturing costs, investment and contract performance. The contracts support the thesis; they do not settle it.

Manufacturing investment can change the ranking

Micron spent $30.712 billion on property, plant and equipment in fiscal 2026. Its annual filing projects approximately $25 billion of capital spending net of government incentives in the first half of fiscal 2027, and more in the second half. That planned acceleration means an investor should not assume today's cash yield survives unchanged even if revenue keeps rising.

Higher capital spending can be productive. New capacity and better manufacturing processes can expand supply, reduce unit costs and support future generations of memory. But the cash leaves before all those benefits arrive. A stock can look cheap on trailing earnings while the next period requires a larger reinvestment commitment. This is why the value case needs both a profit assumption and a capital-spending assumption.

Nvidia's direct capital purchases are much smaller relative to its sales because its business relies on outside manufacturing partners. That does not make it free of capital commitments. Its filing describes substantial supply and capacity obligations, as well as investments and arrangements supporting the wider AI infrastructure ecosystem. Investors should examine these commitments alongside factory spending, rather than awarding Nvidia a risk-free label because it does not own the same manufacturing footprint.

The relevant distinction is where the obligation sits. Micron carries substantial manufacturing investment on its own cash-flow statement. Nvidia can carry exposure through working capital, supplier commitments and investments. A useful comparison follows the cash and obligations across those categories instead of assuming one business model has no financial demands.

Debt and dilution: look beyond the buyback headline

At their respective reporting dates, Nvidia had $33.366 billion of debt at carrying value, compared with $56.586 billion of cash, cash equivalents and marketable debt securities. That leaves approximately $23.220 billion before considering other liabilities. Its equity investments are separate: they can fluctuate and should not automatically be treated like cash when assessing financial flexibility.

Micron reported $5.179 billion of current and long-term debt. Cash and short- and long-term marketable investments totaled $73.453 billion, excluding the small restricted-cash amount in its broader liquidity headline. Subtracting debt gives $68.274 billion. That is not a complete measure of distributable surplus, especially given customer deposits, tax liabilities and planned factory investment. Both companies have substantial resources; neither balance sheet can be reduced to a single cash-minus-debt headline.

Share count also matters more than a board authorization. Nvidia's period-end outstanding shares fell from 24.304 billion in January to 24.147 billion in July, about 0.65%, after repurchases and issuance. Micron's rose from 1.122 billion to 1.131 billion over its fiscal year, about 0.80%. Those intervals differ, so the percentages are not a matched-period buyback competition. They show why gross repurchase dollars do not tell shareholders the net change in their ownership claim.

The correct valuation denominator uses common shares outstanding, while diluted earnings per share incorporates potentially dilutive awards through a weighted-average calculation. Neither figure is the same as the maximum number of shares a company is authorized to issue. Using the wrong share measure can distort an otherwise careful comparison by billions of dollars.

How much of Micron's cash advantage can disappear?

A sensitivity makes the valuation gap tangible. Hold both reconstructed equity values fixed. Keep Nvidia's trailing cash-after-capital-purchases result unchanged, and reduce Micron's result by different amounts. This is deliberately a stress test, not a forecast: it isolates how much deterioration would erase Micron's starting advantage without pretending to know the next memory cycle.

Micron's cash measure would need to fall by approximately 58%, to about $24.8 billion, to match Nvidia's present 2.42% cash yield under those fixed assumptions. That is a substantial cushion in a relative valuation comparison. It does not mean Micron's share price cannot fall sharply before that threshold, nor that a 2.42% yield is an appropriate required return. Investors may demand a higher yield from a more cyclical business even when its current cash generation is stronger.

The earnings version tells a related story. At unchanged prices, cutting Micron's $74.33 trailing diluted EPS in half raises its P/E from 12.17 to about 24.35. That remains below Nvidia's current 27.45 trailing multiple. But Nvidia's future earnings could also grow or decline, and Micron's cash investment needs could rise while earnings fall. A sensitivity is useful precisely because its assumptions are visible; it becomes misleading when those assumptions are presented as predictions.

Which stock has the stronger case now?

For an investor emphasizing current valuation and willing to follow the memory cycle closely, Micron has the more compelling starting numbers. Its cheaper earnings multiple is supported by a higher cash yield under a consistent calculation, rather than being only a product of an optimistic analyst forecast. Customer commitments also offer a specific piece of evidence against assuming an immediate return to old-cycle economics.

For an investor emphasizing platform durability, Nvidia remains the more persuasive business-model choice, provided its premium is accompanied by improved cash conversion. Its integration of compute, networking and software gives customers reasons to stay beyond the availability of one component. The strongest challenge to that argument is financial: slow collections, rising obligations or weaker customer returns could pressure cash even while revenue remains impressive.

Neither conclusion justifies ignoring entry price or concentrating an entire portfolio in AI infrastructure. An investor unable to tolerate a sharp sector drawdown may find neither individual stock suitable. Owning both splits exposure between different economics but retains the shared customer-spending cycle. A broader allocation can be a reasonable alternative without requiring a precise prediction about which semiconductor company wins the next quarter.

The evidence that would change this comparison is concrete. Nvidia's case improves if collections catch up and operating cash flow becomes a stronger proportion of earnings. Micron's case improves if contractual pricing support and higher output fund its investment program without a large decline in cash generation. A deterioration in either test matters more than another headline about the size of the AI market. The next decision should follow those reported numbers, not a refreshed slogan about the better AI stock.