Financial-Economics.nl

Author: Admin

  • Nvidia’s Revenue Is Booming. So Is their Financial Support Around It.

    Nvidia’s Revenue Is Booming. So Is their Financial Support Around It.

    Over the same period, a broad measure of Nvidia’s disclosed financial exposures to customers and suppliers increased much faster than revenue itself.

    That is the most striking conclusion I get from Nvidia’s latest filings.

    Between October 2025 and the latest disclosed figures, Nvidia’s trailing-twelve-month revenue increased from $187 billion to $303 billion, an increase of 62%.

    In this post you’ll read that over the same period, a broad measure of Nvidia’s disclosed customer- and supplier-related financial exposures increased from approximately $94 billion to $559 billion. That is an increase of almost 500%.

    The latest 10-Q contains another $56 billion of newly disclosed customer-facing commitments. Including those brings the broad exposure to $615 billion, that is 2x Nvidia’s trailing annual revenue.

    Given that Nvidia is increasingly acting as an investor, guarantor, and trade-credit provider: How do we keep track of Nvidia’s financial support exposure?

    The Stimulation Index

    I divide Nvidia’s total financial support exposure into two groups:

    1. Customer Stimulation: measures financial exposure that can increase the financial capacity of Nvidia’s customers or the broader demand ecosystem.
    2. Supplier Stimulation: measures Nvidia’s enormous contractual commitments to the supply side.

    Each number below is expressed as a percentage of Nvidia’s trailing-twelve-month revenue.

    Nvidia financial exposure versus trailing twelve-month revenue, October 2025 to July 2026
    Nvidia financial exposure versus trailing twelve-month revenue, October 2025 to July 2026
    Oct. 2025Jan. 2026Apr. 2026Jul. 2026, latest disclosed*
    TTM revenue$187B$216B$254B$303B
    Customer Stimulation Index (A)23.6%38.7%55.4%111.0%
    Supplier Stimulation Index (B)26.9%44.1%46.9%92.1%
    Stimulation Index (A + B)50.5%82.8%102.4%203.1%

    *Q2 FY2027 filing, incl. Aug. subsequent events

    Revenue is growing extraordinarily quickly, but the financial exposure surrounding that revenue is growing much faster.

    So what exactly is inside these numbers?

    Working capital: selling today, collecting later

    Imagine Nvidia sells a customer $1 billion of GPUs. If the customer pays immediately, Nvidia gets its cash and the transaction is finished. But Nvidia does not always require immediate payment.

    Nvidia says that payment is generally due shortly after delivery, but certain investment-grade customers can receive payment terms ranging from 90 days to one year.

    Economically, Nvidia is temporarily financing part of the purchase. You can see this accumulating in accounts receivable. In turn, suppliers temporarily finance part of Nvidia’s purchases, measured in accounts payable. I use the difference as a simple net trade-credit proxy.

    Oct. 2025Jan. 2026Apr. 2026Jul. 2026
    Accounts receivable (A)$33.4B$38.5B$40.7B$63.1B
    Accounts payable (B)$8.6B$9.8B$13.1B$15.1B
    Net trade-credit (A − B)$24.8B$28.7B$27.6B$48.0B
    Net trade-credit / TTM revenue13.2%13.3%10.9%15.8%

    There is another clue in the cash-flow statement. During the first half of FY2027, the increase in accounts receivable absorbed $25 billion of cash. In the first half one year earlier, it absorbed only $5 billion. Nvidia’s commentary explains extended payment terms on large multi-quarter agreements with certain investment-grade customers partially held back operating cash flow.

    If you want to verify this yourself, go to the latest Q2 FY2027 10-Q and look at:

    Note 7 — Supplemental Financial Statement Information, particularly the customer-payment-term discussion;

    and the Condensed Consolidated Statements of Cash Flows, followed by the liquidity discussion in Management’s Discussion and Analysis.

    Guarantees: Nvidia effectively co-signs the financing

    Suppose a company wants to build an enormous data center. A landlord, lender or infrastructure provider asks: “What happens if this customer cannot pay?” Nvidia can step in: If certain obligations are not paid, we will stand behind them. That guarantee can make infrastructure much easier to finance.

    Nvidia already had approximately $3.5 billion of maximum gross exposure from guarantees supporting land, power and shell obligations for AI-cloud partners. Then, in August 2026, Nvidia entered into additional guarantees relating to the SB Energy PORTS-Pike project in Ohio. Those guarantees are capped at $105 billion. Nvidia signed it in August, and the obligations become effective in phases as the relevant facilities become operational. It is an extraordinary amount of contractual credit support.

    Combined, Nvidia presents the following maximum gross guarantee exposure:

    Guarantees

    		Oct. 2025Jan. 2026Apr. 2026Jul. 2026, latest disclosed*
    AI-cloud land, power and shell guarantees (A)$0.9B$3.5B$3.5B$3.5B
    SB Energy guarantees (B)–––$105.0B
    Total guarantees (A + B)$0.9B$3.5B$3.5B$108.5B
    Total guarantees / TTM revenue0.5%1.6%1.4%35.8%

    To verify the numbers, open the latest Q2 FY2027 10-Q and go to:

    Note 10 — Commitments and Contingencies → Additional Commitments and Guarantees.

    Nvidia provides the $108.5 billion maximum gross exposure directly in a table.

    Equity investments: Nvidia becomes an investor in its ecosystem

    Suppose an AI infrastructure company needs billions of dollars to expand. Nvidia does not necessarily have to lend it money or guarantee its debt. It can simply become an investor. That is increasingly what Nvidia is doing.

    At July 26, Nvidia had approximately $99 billion of equity investments and another $25 billion of equity-investment commitments.

    These investments span companies operating inside the broader AI ecosystem around Nvidia. Nvidia is committing an increasingly large amount of its own capital to the ecosystem in which it operates.

    US$BOct. 2025Jan. 2026Apr. 2026Jul. 2026
    Equity investments, carrying value$12.1B$40.0B$82.4B$98.9B
    Equity investments / TTM revenue6.4%18.5%32.5%32.7%

    And Nvidia still had another $25 billion of equity-investment commitments at the latest quarter-end. Nvidia has significant financial exposure to the success of the AI buildout. If valuations or business prospects deteriorate, Nvidia carries part of that financial risk as well.

    To verify the numbers yourself, open the latest Q2 FY2027 10-Q and look at:

    Notes 5 and 6 — marketable/public and non-marketable securities, for the carrying values;

    Note 6 — Equity Method Investments, for investments in infrastructure-financing entities;

    and:

    Note 10 — Commitments and Contingencies, for the outstanding $25 billion of equity-investment commitments.

    Nvidia also summarizes the position directly in Management’s Discussion and Analysis: approximately $99 billion of equity investments and $25 billion of additional equity-investment commitments at July 26.

    Infrastructure commitments: Nvidia becomes an anchor customer

    Imagine an AI-cloud company wants to build a huge new data center. The company may have potential customers, but constructing the facility requires enormous upfront spending on land, power, buildings, servers and networking equipment. Before committing that capital, its financiers want confidence that somebody will actually use the infrastructure. Nvidia can provide part of that confidence by committing to buy cloud capacity itself.

    The latest filing introduces a very large new category of financial support:

    • $36 billion of AI-cloud commitments
    • $20 billion of data-center leases not yet commenced for third parties
    Oct. 2025Jan. 2026Apr. 2026Jul. 2026
    AI-cloud agreements (A)–––$36.0B
    Data-center leases not commenced for third parties (B)–––$20.0B
    Infrastructure commitments (A + B)–––$56.0B
    Infrastructure commitments / TTM revenue———18.5%

    Note 10 — Commitments and Contingencies → Additional Commitments: Look specifically for the table titled “Future commitments by fiscal year as of July 26, 2026.”

    Supplier commitments: Nvidia underpins the other side of the buildout

    Nvidia is making even larger commitments on the supply side. Nvidia expects to need an extraordinary amount of advanced memory, manufacturing capacity and other components over the next several years. Its suppliers face the opposite problem from an AI-cloud operator: before investing billions in new factories and capacity, they want confidence that somebody will buy what those factories produce. A long-term purchasing commitment from Nvidia can provide that confidence.

    And the numbers have become enormous.


    		Oct. 2025Jan. 2026Apr. 2026Jul. 2026
    Supply and capacity commitments$50.3B$95.2B$119.0B$279.0B
    Commitments / TTM revenue26.9%44.1%46.9%92.1%

    Nvidia is making enormous contractual commitments to secure the supply chain it expects to need, although the company notes that some supplier agreements may be cancelable, rescheduled or adjustable before firm orders are placed. But economically, those commitments can still be important to the AI buildout. If Nvidia promises to purchase enormous quantities of memory, manufacturing capacity or other components, suppliers have much greater confidence when deciding whether to expand production. Nvidia’s balance sheet and contractual purchasing power help investment on the supply side of the ecosystem.

    That also creates substantial exposure for Nvidia.

    If demand falls short, Nvidia may find itself committed to far more inventory or capacity than it ultimately needs.

    Note 10 — Commitments and Contingencies → Supply and capacity: The filing provides a schedule totaling $279 billion, including approximately:

    • $92 billion during the remainder of FY2027
    • $87 billion in FY2028
    • $88 billion in FY2029

    Putting it together: Nvidia’s financial exposure to the AI ecosystem

    Put these mechanisms next to each other and Nvidia begins to look like much more than just a semiconductor supplier.

    • It extends payment terms to customers.
    • It guarantees infrastructure obligations.
    • It invests equity capital into companies operating across the ecosystem.
    • It commits to purchasing cloud and data-center capacity.
    • And it enters enormous long-term purchasing commitments with suppliers.

    At the latest disclosed values, the customer-side exposure looks like this:

    Customer-side componentExposure% TTM revenue
    Net trade-credit proxy$48.0B15.8%
    Maximum gross guarantees$108.5B35.8%
    Equity investments, carrying value$98.9B32.7%
    Equity-investment commitments$25.0B8.3%
    New customer-facing infrastructure commitments$56.0B18.5%
    Customer Stimulation Exposure$336.4B111.0%

    Then add another $279 billion of supplier commitments, equivalent to 92.1% of trailing revenue.

    The fully disclosed latest exposure therefore becomes:

    $615.4B of broadly defined ecosystem-related assets, commitments, guarantees and net working-capital exposure, equivalent to 203.1% of trailing revenue.

    This amount captures the scale of Nvidia’s assets, commitments, credit support and working-capital exposure tied to the continued expansion of the AI ecosystem.

    The question to track each new quarter is:

    How much more financial support is Nvidia placing behind the AI buildout, and is that exposure continuing to grow faster than the business itself?

    • Financial Times / Alphaville: “Nvidia’s $200bn ‘balance sheet-as-a-service’” Read at the Financial Times
    • Linas Beliūnas: “Brilliant: the infinite AI money glitch explained” Read on LinkedIn
    • Pat Flynn: “What could go wrong? OpenAI buys $300 billion compute power from Oracle; Nvidia invests up to $100 billion in OpenAI Read on LinkedIn

  • High Bandwidth Memory (HBM): Why AI Is Driving Memory Stocks

    For decades, memory-chip manufacturers were viewed as highly cyclical commodity businesses. Profits rose when supply was tight, collapsed when manufacturers added too much capacity, and eventually recovered when demand caught up.

    Artificial intelligence may be changing that pattern.

    Advanced AI accelerators require enormous quantities of extremely fast memory. This has turned High Bandwidth Memory, or HBM, into one of the most important and constrained components in the AI supply chain.

    Investors are therefore asking a new question:

    Has AI permanently improved the economics of the memory industry, or is this simply another unusually strong cycle?

    This article examines five developments behind the growing interest in memory stocks:

    1. Advanced memory has become essential for advanced AI systems.
    2. Supply remains highly constrained.
    3. AI infrastructure spending is enormous.
    4. Memory technology may become less commoditized.
    5. Memory-stock valuations still reflect considerable skepticism.

    What Is HBM?

    GPUs perform the calculations required to train and run large AI models. However, a GPU can only work efficiently when data reaches its computing cores fast enough.

    That is where memory becomes critical.

    AI workloads constantly move large quantities of data, including:

    • Model weights
    • Activations
    • Gradients
    • Optimizer states
    • Training data
    • Intermediate results

    When memory cannot deliver this data quickly enough, expensive computing cores sit idle.

    HBM is designed to reduce that bottleneck.

    Key Benefits of HBM

    High Bandwidth Memory offers:

    • Extremely high data-transfer rates
    • Greater energy efficiency per transferred bit
    • High memory capacity close to the processor
    • A compact physical footprint
    • Better performance for bandwidth-intensive workloads

    HBM is placed next to the GPU or accelerator and built from multiple vertically stacked memory dies. These dies are connected using through-silicon vias, or TSVs, which allow data to move through the stack at very high speed.

    This architecture is more complex and expensive than conventional memory, but it provides the bandwidth required by modern AI systems.

    HBM vs. GDDR

    GDDR, or Graphics Double Data Rate memory, is the high-speed memory used in most consumer graphics cards.

    Both GDDR and HBM are designed for demanding graphics and computing workloads, but they are optimized differently.

    FeatureGDDRHBM
    Physical layoutMemory chips placed around the processorMemory dies stacked vertically beside the processor
    BandwidthHighExtremely high
    Energy efficiencyModerateHigher
    Manufacturing costLowerHigher
    Packaging complexityLowerMuch higher
    Common applicationsGaming GPUs and workstationsAI accelerators, supercomputers and data centers

    GDDR remains an excellent choice for gaming, workstations and smaller AI workloads. HBM is preferred when bandwidth and power efficiency are more important than cost.

    Approximate Bandwidth by Memory Type

    Memory typeTypical aggregate bandwidth
    GDDR6Approximately 500–1,000 GB/s
    GDDR7Approximately 1–2 TB/s, depending on configuration
    HBM3Approximately 3–5 TB/s
    HBM3EApproximately 5–8+ TB/s

    Exact performance varies by product and memory configuration, but the broader point is clear: HBM can provide several times the bandwidth of conventional graphics memory.

    Why AI Accelerators Use HBM

    Most advanced AI-training accelerators rely on HBM.

    Examples include:

    • NVIDIA H100
    • NVIDIA H200
    • NVIDIA Blackwell accelerators
    • AMD Instinct MI300X
    • AMD MI350-series accelerators
    • Google Tensor Processing Units
    • Amazon Web Services Trainium processors

    These systems are designed to train large language models and other computationally intensive AI applications.

    A faster processor does not automatically solve the memory problem. In many cases, it makes the problem more severe because the processor consumes data more quickly.

    This creates an important relationship:

    As AI computing performance increases, memory bandwidth must usually increase with it.

    That relationship has made HBM a strategic part of accelerator design rather than a secondary component.

    What Happens Without HBM?

    Companies that cannot obtain enough HBM have several alternatives, but all involve trade-offs.

    Use GDDR6 or GDDR7

    GDDR is more widely available and easier to integrate. However, matching HBM performance may require more chips, more board space and more power.

    Use DDR5 or LPDDR

    Conventional server or low-power memory may work for less demanding inference and edge applications. It generally lacks the bandwidth required for frontier-model training.

    Add More On-Chip Memory

    Chip designers can increase SRAM and cache capacity, allowing frequently used data to remain closer to the computing cores. However, SRAM is expensive and consumes substantial chip area.

    Improve Software Efficiency

    Quantization, pruning, sparsity, data compression and better scheduling can reduce memory traffic. These techniques improve efficiency but do not fully replace HBM in the most bandwidth-intensive workloads.

    Buy Complete Accelerator Systems

    Instead of designing a custom processor, companies can purchase GPUs or accelerator modules that already include HBM. This simplifies development but increases dependence on established hardware suppliers.

    For advanced AI training, the trade-off is usually straightforward: without HBM, systems tend to offer lower performance, consume more power, occupy more space or rely more heavily on incumbent vendors.

    Why HBM Supply Is So Tight

    Strong demand alone does not create attractive industry economics. The more important issue is that supply has struggled to keep up.

    Only three companies currently manufacture leading-edge HBM at significant scale:

    • SK Hynix
    • Samsung Electronics
    • Micron Technology

    This concentrated supplier base gives the industry a structure that is unusual for a large and strategically important technology market.

    HBM Is Difficult to Manufacture

    Producing stacked memory requires several complex processes:

    • Advanced DRAM fabrication
    • Precise thinning of memory dies
    • Vertical stacking of multiple dies
    • Through-silicon via connections
    • Advanced packaging
    • Thermal management
    • Extensive testing and customer qualification

    Every additional layer increases manufacturing difficulty.

    A defect in one component can reduce the yield of the entire stack. Manufacturers must therefore achieve high levels of precision across both memory production and packaging.

    These requirements create meaningful barriers to entry.

    Capacity Is Often Committed in Advance

    AI-chip manufacturers and cloud providers need predictable access to memory. As a result, much of the industry’s near-term HBM output is secured through customer agreements before it is produced.

    When most capacity is already allocated, customers have limited ability to obtain additional supply at short notice.

    This gives established manufacturers greater visibility and potentially stronger pricing power than they have historically enjoyed in conventional memory markets.

    Capacity Takes Time to Expand

    Increasing advanced memory output involves more than producing additional DRAM wafers.

    Manufacturers may need to:

    • Add leading-edge wafer capacity
    • Expand stacking and packaging facilities
    • Purchase specialized equipment
    • Improve production yields
    • Qualify new products with customers
    • Coordinate designs with accelerator manufacturers

    These investments can take months or years to produce meaningful output.

    That delay allows demand to outpace supply, even while manufacturers invest aggressively in expansion.

    AI Infrastructure Spending Is Enormous

    Microsoft, Amazon, Meta Platforms, Alphabet and other large technology companies are spending heavily on data centers, accelerators, networking equipment and supporting infrastructure.

    Only part of this spending goes directly to memory. Nevertheless, HBM captures significant economic value because it is both expensive and essential.

    Industry estimates have suggested that the global HBM market could grow from tens of billions of dollars annually to a substantially larger market later in the decade.

    This memory technology can also represent a meaningful portion of the manufacturing cost of a high-end AI accelerator.

    This helps explain why memory suppliers are increasingly viewed as major beneficiaries of AI investment rather than merely component vendors.

    The investment argument is not that all AI spending flows to HBM. It is that nearly every advanced accelerator requires it, and the number of HBM stacks used per system continues to increase.

    Could the HBM Trend Reverse?

    Demand for this technology will not necessarily grow forever.

    Several developments could reduce the amount of premium memory required per unit of AI computation.

    More Efficient Models

    Better quantization, pruning, sparsity and model architectures could reduce memory usage and data movement.

    Larger Caches and More On-Chip Memory

    Future processors may keep more frequently accessed information close to the computing cores, reducing repeated HBM access.

    Better Software

    Compilers, scheduling systems and memory-management software could improve utilization and reduce unnecessary data transfers.

    New Computing Architectures

    Near-memory computing, processing-in-memory, optical interconnects or new memory technologies could eventually reduce dependence on today’s HBM architecture.

    Slower AI Infrastructure Spending

    HBM demand is closely tied to accelerator deployments. If cloud providers slow their capital spending after the current build-out, memory demand could weaken.

    Faster-Than-Expected Supply Expansion

    Improved yields, better packaging equipment and aggressive investment could allow supply to catch up with demand sooner than investors expect.

    The key question is not whether AI systems will become more efficient. They almost certainly will.

    The question is whether efficiency improvements will reduce memory demand faster than models, datasets and computing systems grow.

    Historically, greater computing efficiency has often led to more usage rather than less.

    Is This Memory Cycle Different?

    Traditional memory markets have followed a familiar pattern:

    1. Demand rises.
    2. Supply becomes scarce.
    3. Prices and profits increase.
    4. Manufacturers invest in new capacity.
    5. Supply eventually exceeds demand.
    6. Prices and profits decline.

    The current debate is whether HBM changes that cycle.

    The Bullish View

    Supporters of the structural-growth thesis argue that:

    • AI demand is long term rather than temporary.
    • HBM is more specialized than conventional DRAM.
    • Manufacturing and packaging are technically difficult.
    • Customer qualification creates switching costs.
    • Capacity expansion remains slow.
    • Longer-term agreements improve revenue visibility.
    • Higher barriers to entry could support stronger margins.

    The Skeptical View

    Skeptics argue that:

    • Memory has always been cyclical.
    • High profits encourage aggressive capital spending.
    • Production yields improve over time.
    • Samsung and Micron may gain market share.
    • Customers will search for cheaper alternatives.
    • AI capital spending may eventually slow.
    • Today’s shortages could become tomorrow’s oversupply.

    Both views can be partly correct.

    HBM may make memory more specialized and less commoditized without eliminating cyclicality altogether.

    Traditional Memory vs. AI Memory

    Traditional memoryAdvanced AI memory
    Highly standardizedMore specialized
    Frequently sold through shorter-term marketsMore capacity committed in advance
    Easier customer substitutionLengthy customer qualification
    Broad end marketsConcentrated AI and data-center demand
    Lower packaging complexityAdvanced stacking and packaging
    Strongly commodity-drivenGreater technical differentiation

    The most likely outcome may be a memory industry that remains cyclical but experiences higher barriers to entry, longer product cycles and stronger profitability than in the past.

    Future Memory Technologies

    Memory manufacturers are working on technologies that make HBM faster, larger and more energy efficient.

    Major areas of development include:

    • HBM4 and later generations
    • Taller memory stacks
    • Higher-capacity dies
    • Hybrid bonding
    • Improved thermal management
    • Custom HBM designed for specific accelerators
    • More efficient packaging
    • Faster interfaces between memory and compute

    These developments require close cooperation among memory suppliers, chip designers, foundries and packaging companies.

    That collaboration may strengthen relationships between customers and suppliers. It also makes it more difficult for new competitors to enter quickly.

    However, manufacturing challenges are not permanent. As processes improve and yields rise, production can expand and costs can fall.

    Technological complexity supports pricing power—but it does not guarantee it indefinitely.

    Key Risks to the Investment Thesis

    Three risks matter most.

    1. AI Becomes Less Memory-Intensive

    More efficient models and software could reduce the amount of HBM required for each accelerator or workload.

    2. Supply Expands Faster Than Demand

    Aggressive investment, improving yields and more qualified suppliers could turn today’s shortage into excess capacity.

    3. AI Spending Slows

    If hyperscalers pause or reduce infrastructure investment, demand for accelerators and HBM could weaken sharply.

    Additional risks include:

    • Alternative memory technologies
    • Processing-in-memory architectures
    • Faster packaging throughput
    • Increased competition
    • Geopolitical disruption
    • Customer concentration
    • Pricing pressure from major accelerator vendors

    These risks are important because the market may be valuing memory companies on unusually high current earnings.

    The Memory Stocks Investors Are Watching

    Three companies dominate the HBM discussion.

    SK Hynix

    SK Hynix is widely regarded as the current HBM leader and a major supplier to the AI-accelerator market.

    Its early position in advanced HBM has allowed it to benefit substantially from growing demand. Investors are watching whether it can maintain its technology lead as competitors increase production.

    Micron Technology

    Micron is expanding its HBM business and has benefited from improving memory demand, higher-value products and stronger margins.

    The central question is whether Micron can scale advanced HBM production while maintaining attractive yields and profitability.

    Samsung Electronics

    Samsung remains one of the world’s largest memory manufacturers and is investing heavily in advanced HBM.

    Its scale, manufacturing capabilities and financial resources make it a formidable competitor. Investors are watching how quickly Samsung can qualify new products and win a larger share of next-generation accelerator programs.

    Why Memory-Stock Valuations Can Appear Unusual

    Despite strong earnings growth and significant share-price gains, memory companies can trade at modest-looking earnings multiples.

    That does not necessarily mean the stocks are inexpensive.

    In highly cyclical industries, earnings are often strongest near the top of the cycle. A low price-to-earnings ratio may therefore indicate that investors expect profits to fall.

    This creates the central valuation debate.

    The Bearish Interpretation

    Current earnings may represent a cyclical peak. As new capacity arrives, prices and margins could normalize.

    The Bullish Interpretation

    HBM may support structurally higher margins, longer customer agreements and more durable profits than conventional memory.

    Investors are therefore not only estimating future HBM demand. They are also deciding how much of today’s profitability should be treated as sustainable.

    That distinction can have a major effect on valuation.

    What Investors Should Monitor

    Rather than focusing only on headline AI spending, investors may want to track:

    • HBM production capacity
    • Customer qualification announcements
    • Manufacturing yields
    • HBM pricing
    • Packaging capacity
    • Capital-expenditure plans
    • Market-share changes
    • Memory content per accelerator
    • Hyperscaler AI spending
    • Accelerator shipment growth
    • New memory architectures
    • Gross-margin trends

    These indicators can help reveal whether the industry is experiencing a durable structural change or approaching another cyclical peak.

    Bottom Line

    The enthusiasm surrounding memory stocks is no longer primarily about PCs or smartphones.

    It is about the possibility that memory—especially HBM—has become a strategic bottleneck in the AI supply chain.

    Advanced AI accelerators require enormous bandwidth. Only a small number of companies can currently manufacture leading-edge HBM at scale, and expanding supply is technically difficult, expensive and slow.

    That combination has given memory suppliers a degree of pricing power and customer visibility that has historically been rare in the industry.

    However, the familiar risks have not disappeared.

    High profits encourage investment. Manufacturing yields improve. Competitors catch up. Customers search for alternatives. Demand can slow.

    HBM may therefore make memory less commoditized without making it non-cyclical.

    The investment question is not whether this technology matters. It clearly does.

    The real question is whether its technical complexity and strategic importance will allow memory manufacturers to earn higher and more durable returns—or whether the industry’s traditional boom-and-bust cycle will eventually reassert itself.

    Sources and Further Reading

    Disclosure: This article is for informational purposes only and does not constitute investment advice. Market estimates, valuations and product specifications can change. Readers should verify current figures before making investment decisions.

  • S&P 500 Without Magnificent 7: A Rebalanced Index Experiment

    S&P 500 Without Magnificent 7: A Rebalanced Index Experiment

    Introduction

    What happens to the S&P 500 without Magnificent 7 stocks if those companies are removed and the remaining constituents are rebalanced back to 100%?

    Since the start of 2023, the S&P 500 has delivered a very strong price return. But a large part of that performance has been driven by a small group of mega-cap technology and growth companies: Nvidia (NVDA), Apple (AAPL), Amazon (AMZN), Microsoft (MSFT), Alphabet Class A (GOOGL), Alphabet Class C (GOOG), Meta Platforms (META) and Tesla (TSLA).

    In this post, I use open-spx to create a rebalanced synthetic version of the S&P 500 without Magnificent 7 stocks. The goal is educational: to better understand how much these companies contributed, how concentrated the index became, and how the rest of the market performed when treated as a fully invested portfolio.

    This is not investment advice, and it is not a recommendation to buy or sell any stock, index product or ETF. The analysis is based on price returns only, not total returns, so dividends are not included.

    The setup

    The previous post on this site explained how open-spx can be used to estimate S&P 500 return contributions from constituent-level prices and inferred weights. In this post, I extend that idea by creating a synthetic index.

    The S&P 500 without Magnificent 7 experiment is a counterfactual index construction exercise. It does not ask whether the excluded companies are good or bad investments. Instead, it asks a narrower question: how would the S&P 500 have behaved if those stocks were removed and the remaining constituents were scaled back to a full 100% allocation?

    The workflow is:

    1. Run a regular S&P 500 replication.
    2. Estimate the constituent weights through time.
    3. Remove the excluded tickers.
    4. Rebalance the remaining weights back to 100%.
    5. Compute a new synthetic price-return series.
    6. Reuse the same contribution analysis on the residual index.

    For this experiment, the excluded tickers are:

    • NVDA — Nvidia
    • AAPL — Apple
    • AMZN — Amazon
    • MSFT — Microsoft
    • GOOGL — Alphabet Class A
    • GOOG — Alphabet Class C
    • META — Meta Platforms
    • TSLA — Tesla

    There are eight tickers because Alphabet appears through two share classes: Alphabet Class A (GOOGL) and Alphabet Class C (GOOG). In the charts these are shown separately.

    The displayed period is 2023-01-04 to 2026-06-15, using the first available trading observation in the dataset.

    A simplified version of the command is:

    open-spx --start 2023-01-01 \
  --end 2026-06-15 \
  --exclude-tickers NVDA,AAPL,AMZN,MSFT,GOOGL,GOOG,META,TSLA \
  --synthetic-name ex_mag7

    The requested start date is 2023-01-01, but the first available trading date in this run is 2023-01-04.

    How the synthetic ex-Magnificent 7 index is constructed

    The construction is straightforward.

    At every point in time, open-spx starts with the inferred S&P 500 constituent weights. It then removes the excluded tickers and rescales the remaining weights so that the residual universe sums to 100%.

    The formula is:

    new weight = original weight / sum of remaining weights

    For example, if the Magnificent 7 represent 36% of the index on a given date, the remaining stocks represent 64%. The residual stocks are then scaled by:

    1 / 0.64 = 1.56x

    This is the key mechanical detail. The synthetic index is not the S&P 500 minus the Magnificent 7 with the removed weight held in cash. Instead, the removed weight is reallocated to the remaining S&P 500 constituents.

    That distinction is essential for interpreting the results.

    S&P 500 without Magnificent 7 versus the full index

    S&P 500 without Magnificent 7 compared with the full S&P 500 from 2023 to 2026
    The S&P 500 price index and replicated S&P 500 both returned about 96% from 2023-01-04 to 2026-06-15. The rebalanced ex-Magnificent 7 synthetic index returned about 61%.

    The first chart compares three series, all normalized to 100 at the start:

    • the S&P 500 price index,
    • the replicated S&P 500,
    • and the rebalanced ex-Magnificent 7 synthetic index.

    From 2023-01-04 to 2026-06-15, the S&P 500 price index increased by about 96%. The replicated S&P 500 closely tracks it, also ending around 96%.

    The ex-Magnificent 7 synthetic index returned about 61%.

    That is a lower return, but it is still a strong positive result. The main message is not that the Magnificent 7 did not matter. They clearly did. The message is that the rest of the S&P 500 still delivered meaningful returns when treated as a fully invested portfolio.

    In other words, the S&P 500 without Magnificent 7 lagged the full index, but it did not collapse. The residual market still compounded strongly.

    How much did the Magnificent 7 contribute to the S&P 500?

    Nvidia, Apple, Amazon, Microsoft, Alphabet Class A, Alphabet Class C, Meta Platforms, and Tesla contributed about 37 percentage points to S&P 500 returns.
    The excluded Magnificent 7 tickers contributed about 37 percentage points to the S&P 500 price return. Nvidia was the largest contributor, followed by Apple, Amazon, Microsoft, Alphabet Class A, Alphabet Class C, Meta Platforms, and Tesla.

    The excluded Magnificent 7 tickers contributed about 37 percentage points to the S&P 500 price return over this period.

    Nvidia (NVDA) was the largest contributor, adding about 11 percentage points. Apple (AAPL) contributed about 6 percentage points. Amazon (AMZN) and Microsoft (MSFT) each contributed about 4 percentage points. Alphabet Class A (GOOGL), Alphabet Class C (GOOG), Meta Platforms (META) and Tesla (TSLA) each contributed about 3 percentage points when rounded to whole percentages.

    Together, these excluded tickers contributed about 37 percentage points. Relative to the S&P 500’s approximately 96% price return, that means the Magnificent 7 represented roughly 39% of the index return in this run.

    That is a large number. It confirms that recent S&P 500 performance was highly dependent on a small group of mega-cap companies.

    It also explains why the S&P 500 without Magnificent 7 performed differently from the original index. The excluded companies represented a large share of total return contribution.

    Why the S&P 500 without Magnificent 7 is not just subtraction

    Return bridge showing the S&P 500 returned about 96%, the ex-Magnificent 7 synthetic index returned about 61%, and the return reduction was about 35 percentage points.
    Removing the Magnificent 7 reduced the return by about 35 percentage points, less than their raw contribution of about 37 percentage points, because the removed weights were reallocated to the remaining S&P 500 stocks.

    The bridge chart shows the central result of the experiment:

    • S&P 500 cumulative price return: about 96%
    • Ex-Magnificent 7 synthetic return: about 61%
    • Raw excluded contribution: about 37 percentage points
    • Return reduction from exclusion: about 35 percentage points

    At first glance, this might look surprising. If the Magnificent 7 contributed 37 percentage points, why does the synthetic index lag by 35 percentage points?

    The answer is rebalancing.

    This is the most important mechanical detail in the S&P 500 without Magnificent 7 analysis. The excluded weight is not moved to cash. It is reallocated across the residual S&P 500 universe.

    When the Magnificent 7 are removed, their weight is redistributed across the remaining constituents. Those companies receive a larger portfolio weight than they had in the original capitalization-weighted index.

    So the ex-Magnificent 7 index is not simply:

    S&P 500 return - Magnificent 7 contribution

    Instead, it is:

    S&P 500 without Magnificent 7, with the remaining stocks scaled back to 100%

    That is why the return reduction is smaller than the raw contribution of the excluded stocks.

    The contribution did not arrive in a straight line

    Total Magnificent 7 contribution to the S&P 500 rose over time, reaching about 37 percentage points, with Nvidia contributing the largest share.
    The Magnificent 7 contribution did not arrive in a straight line. Total contribution rose strongly from 2023 through 2026, with Nvidia standing out as the largest individual contributor.

    The Magnificent 7 contribution was not smooth.

    The cumulative contribution rose strongly in 2023 and 2024, pulled back at several points, and then continued to rise into 2025 and 2026. Nvidia (NVDA) stands out as the largest individual contributor, while the other names contributed more gradually.

    This matters because concentration risk is not only about the final number. It is also about the path.

    A portfolio that becomes increasingly dependent on a narrow set of companies can perform very well when those companies lead. But it may also become more vulnerable if leadership reverses.

    Concentration increased over time

    The inferred weight of the Magnificent 7 in the replicated S&P 500 rose from about 20% to more than 30%, showing increased index concentration.
    The inferred Magnificent 7 weight increased from roughly one-fifth of the index to around one-third of the replicated S&P 500, highlighting the concentration risk inside a market-cap-weighted index.

    The inferred combined weight of the excluded Magnificent 7 tickers increased significantly over the period.

    At the start of the sample, the excluded group represented roughly one-fifth of the replicated S&P 500. By later in the period, it represented roughly one-third of the index. At some points, the excluded group represented more than one-third of the replicated S&P 500.

    This is the core concentration-risk issue.

    The S&P 500 contains hundreds of companies, but it is market-cap weighted. When a small group of companies becomes very large, a broad-market index can become increasingly exposed to the same few names.

    That does not make those companies bad investments. Nvidia (NVDA), Apple (AAPL), Amazon (AMZN), Microsoft (MSFT), Alphabet (GOOGL and GOOG), Meta Platforms (META) and Tesla (TSLA) are major companies with major economic importance. But it does mean that investors using the S&P 500 as broad-market exposure should understand how concentrated that exposure can become.

    A simple rebalance example

    Example showing Magnificent 7 removal and residual S&P 500 rebalancing, where the remaining stocks are scaled from about 64% to 100%.
    After the excluded Magnificent 7 weight is removed, the residual S&P 500 universe is scaled back to 100%. In this example, the remaining weights are multiplied by about 1.55x.

    The rebalance mechanics chart shows a point-in-time example.

    On the selected date, the excluded Magnificent 7 tickers represented about 36% of the index. The residual universe represented about 64%. After removing the excluded tickers, the remaining weights were scaled back to 100%.

    In this example, the residual weights were multiplied by about 1.55x.

    This is the mechanical reason why the ex-Magnificent 7 index can still compound strongly. The remaining companies are not left at their original reduced weights. They become the whole portfolio.

    Relative performance: when did the ex-Magnificent 7 index lag?

    The rebalanced ex-Magnificent 7 synthetic index generally lagged the S&P 500, although rolling return differences varied over time.
    The ex-Magnificent 7 synthetic index lagged the S&P 500 over the full period, but relative performance varied across shorter rolling windows.

    The ex-Magnificent 7 synthetic index lagged the S&P 500 over the full period, but relative performance varied across shorter windows.

    The cumulative excess return line was mostly negative, meaning the synthetic index generally trailed the original S&P 500. However, the rolling return differences show that this underperformance was not constant.

    There were periods where the ex-Magnificent 7 index performed closer to the S&P 500, and some shorter windows where it performed better.

    This is a useful reminder that market leadership changes over time. The Magnificent 7 dominated much of the period, but not every month and not every quarter looked the same.

    Did removing the Magnificent 7 reduce drawdowns?

    Drawdown comparison of the S&P 500 and the rebalanced ex-Magnificent 7 synthetic index from 2023 to 2026.
    The ex-Magnificent 7 index reduces mega-cap concentration, but its drawdowns were not always smaller than the original S&P 500 over this period.

    A common assumption is that reducing concentration should automatically reduce risk. The drawdown comparison is more nuanced.

    The ex-Magnificent 7 synthetic index does reduce exposure to mega-cap technology and growth stocks. However, its realized drawdowns were not always smaller than the original S&P 500 over this sample. In some periods, the ex-Magnificent 7 index had comparable or even larger drawdowns.

    This is important for interpretation.

    Lower concentration does not guarantee lower realized volatility or smaller drawdowns in every market environment. It reduces one type of risk: dependence on a small group of mega-cap companies. But the residual universe has its own risks.

    A less concentrated portfolio can still decline.

    What drove the residual index?

    Broadcom, Micron Technology, Eli Lilly, Advanced Micro Devices, and Walmart were among the top contributors after removing the Magnificent 7.
    After removing the Magnificent 7 and rebalancing the residual index, Broadcom, Micron Technology, Eli Lilly, Advanced Micro Devices, and Walmart were among the largest contributors.

    After removing and rebalancing away from the Magnificent 7 the top residual contributors were led by Broadcom (AVGO), Micron Technology (MU), Eli Lilly (LLY), Advanced Micro Devices (AMD) and Walmart (WMT).

    Broadcom (AVGO) was the largest residual contributor, followed by Micron Technology (MU). Eli Lilly (LLY) and Advanced Micro Devices (AMD) also made large positive contributions. Walmart (WMT), JPMorgan Chase (JPM), Intel (INTC), RTX (RTX), Oracle (ORCL) and Lam Research (LRCX) were also among the important contributors.

    This is one of the more interesting parts of the experiment.

    When the Magnificent 7 are removed, the residual S&P 500 is not empty. Other large companies and sectors still contribute meaningfully.

    The ex-Magnificent 7 result is therefore not a “no growth” portfolio. It still includes semiconductor companies, healthcare companies, financials, industrials, retailers and other large businesses.

    Which companies detracted?

    Pfizer, Marsh & McLennan, UnitedHealth Group, Nike, and Moderna were among the largest detractors after removing the Magnificent 7.
    The largest detractors in the residual index included Pfizer, Marsh & McLennan, UnitedHealth Group, Nike, Moderna, United Parcel Service, PepsiCo, Estée Lauder, Bristol Myers Squibb, and MSCI.

    The largest residual detractors included Pfizer (PFE), Marsh & McLennan (MMC), UnitedHealth Group (UNH), Nike (NKE), Moderna (MRNA), United Parcel Service (UPS), PepsiCo (PEP), Estée Lauder (EL), Bristol Myers Squibb (BMY) and MSCI (MSCI).

    These detractors were much smaller in absolute contribution than the largest positive contributors. For example, the largest negative contribution in the residual index was roughly half a percentage point.

    This asymmetry is useful. The residual index’s positive return was not driven by an absence of losers. There were still detractors. But the positive contributors outweighed them.

    What this means for investors

    The experiment suggests three main takeaways.

    First, the Magnificent 7 mattered enormously. Over this period, Nvidia (NVDA), Apple (AAPL), Amazon (AMZN), Microsoft (MSFT), Alphabet Class A (GOOGL), Alphabet Class C (GOOG), Meta Platforms (META) and Tesla (TSLA) together contributed about 37 percentage points to the S&P 500.

    Second, the rest of the market still performed well. After removing those tickers and rebalancing the remaining S&P 500 constituents, the synthetic index returned about 61%.

    Third, concentration risk is real, but it should be described carefully. Removing the Magnificent 7 would have reduced exposure to the dominant mega-cap stocks, but it would also have reduced returns over this period.

    The benefit is not that the ex-Magnificent 7 index was better in hindsight. It was not. The benefit is that it was less dependent on a small set of companies.

    That may be valuable in scenarios where:

    • market leadership broadens,
    • mega-cap valuations compress,
    • technology leadership weakens,
    • or investors want less dependence on the same few names.

    But it may hurt in scenarios where the Magnificent 7 continue to dominate index returns.

    This is why the experiment is best understood as an educational concentration-risk study, not as a trading recommendation.

    Limitations

    There are several important limitations.

    First, this is a price-return analysis. Dividends are not included.

    Second, the weights are inferred. They are not official S&P Dow Jones Indices constituent weights.

    Third, the analysis depends on data quality, ticker mapping, corporate actions and the available constituent history.

    Fourth, this is a counterfactual index experiment. It does not include trading costs, taxes, slippage, liquidity constraints or product implementation details.

    Fifth, the result depends on the chosen date range. A different starting point or ending point could produce different conclusions.

    Finally, Alphabet Class A (GOOGL) and Alphabet Class C (GOOG) are treated separately because they are separate listed tickers in the dataset, even though they represent the same company.

    Conclusion

    From 2023-01-04 through 2026-06-15, the S&P 500 price index returned about 96% in this run. The replicated S&P 500 closely matched that result. The Magnificent 7 contributed about 37 percentage points, or roughly 39% of the index return.

    When those tickers were removed and the remaining S&P 500 constituents were rebalanced back to 100%, the synthetic ex-Magnificent 7 index still returned about 61%.

    That is the key result.

    The Magnificent 7 were extremely important. But the rest of the S&P 500 was not irrelevant. A rebalanced residual universe still produced a strong return.

    The S&P 500 without Magnificent 7 is therefore best understood as a concentration-risk experiment. It shows how much the index depended on a small group of companies, while also showing that the residual market still delivered meaningful returns.

    For investors, the lesson is not that one version is obviously better than the other. The lesson is that index construction matters. A market-cap-weighted index can become highly concentrated, and understanding that concentration is essential when interpreting broad-market returns.

    Related links

    The S&P 500 is maintained by S&P Dow Jones Indices.

    This experiment uses price-index style analysis rather than official total-return index methodology. More information about S&P Dow Jones Indices methodology is available from S&P Dow Jones Indices methodology resources.

    The code used for this experiment is available in the open-spx project on GitHub.