AI Bubble Statistics 2026
By Axis Intelligence Research
Co-authors: Sarah Mitchell, AI & Machine Learning, and Sarah Davis, Finance & Digital Finance | Last updated: August 5, 2026 | License: CC BY 4.0
Alphabet and Meta spent $76.0 billion building AI capacity in the June 2026 quarter. Their operations produced $70.9 billion. According to Axis Intelligence Research, that leaves 93.3 cents of self-generated cash per dollar of buildout, against $1.35 a year earlier — the gap is now funded by issuance.
Quick Answer
The Axis Intelligence Research Capex Funding Self-Sufficiency Ratio (CFSR™) reads 93.3 as of June 30, 2026, meaning the two hyperscalers that publish full quarterly cash flow detail funded 93.3% of their combined capital expenditure from operations. A year earlier the same calculation produced 135.1. Alphabet reported negative free cash flow of $5.86 billion in Q2 2026, and Meta’s free cash flow fell to $784 million from $8.55 billion. Meanwhile, 19.8% of U.S. businesses reported using AI as of May 3, 2026.
Key Findings
- According to Axis Intelligence Research, the combined Capex Funding Self-Sufficiency Ratio for Alphabet and Meta fell 41.8 points year over year, from 135.1 in Q2 2025 to 93.3 in Q2 2026.
- Alphabet reported free cash flow of negative $5.86 billion for the quarter ended June 30, 2026, against $24.46 billion in the quarter ended September 30, 2025, per its Q2 2026 earnings release.
- According to Axis Intelligence Research, Alphabet and Meta raised $99.3 billion of external equity and debt capital in Q2 2026 alone — 130.7% of the $76.0 billion they spent on capital expenditure that quarter.
- Large U.S. banks held about $450 billion of committed commercial and industrial exposure to AI-adjacent industries in late 2025, roughly 13% of total C&I commitments, up from about $250 billion and 9% in 2015, according to the Federal Reserve Bank of Chicago.
- The U.S. Census Bureau reported an AI use rate of 19.8% among all U.S. businesses as of May 3, 2026, against 39.7% in the Information sector and about 14% in Retail Trade.
What Does the AI Bubble Data Actually Show in 2026?
The phrase “AI bubble” is doing two jobs at once, and separating them is the first analytical task. One claim is about prices — that equity valuations for AI-exposed companies embed growth that will not arrive. The other is about funding structure — that the buildout is increasingly financed by debt and equity issuance rather than by the cash the underlying businesses generate. The first claim is hard to settle before the fact. The second is visible in filings, quarter by quarter, and it is where this dataset concentrates.
On prices, the Federal Reserve’s May 2026 Financial Stability Report is the cleanest institutional reading available. Reflecting market conditions as of April 23, 2026, the report describes asset valuation pressures as elevated: the forward price-to-earnings ratio for S&P 500 firms remained well above its historical median of 16.00, and the Board’s estimate of the equity premium — the extra return investors demand for holding stocks over Treasuries — sat near a twenty-year low against a historical median of 4.59 percentage points. Public equities outstanding stood at $83.1 trillion at the end of 2025, having grown 17.6% over the year against a 1997–2025 average annual growth rate of 9.9%.
On funding structure, the same report notes that bond issuance among the largest investment-grade firms involved in cloud computing approached $100 billion in the first quarter of 2026, and that spreads for speculative-grade technology firms widened more than the broader market. Those two sentences describe the mechanism this article measures.
Sarah Davis: The market has spent two years arguing about the multiple. The more useful question is where the money is coming from. A company can carry an expensive multiple for a long time if its own operations pay for the capital program — that is a rich stock, not a fragile one. It becomes fragile when the program is funded by issuance, because issuance depends on the window staying open, and windows close on somebody else’s schedule. The spread widening on speculative-grade tech paper is the first sentence of that story, not the last.
How Much of AI Capex Is Now Funded by Cash Flow? The CFSR™
CFSR™ — Capex Funding Self-Sufficiency Ratio
CFSR™ is an original Axis Intelligence Research metric. It answers one question: for every dollar a hyperscaler spends on capital expenditure in a quarter, how many cents did the business itself generate?
Formula
CFSR = (Net cash provided by operating activities
÷ Total capital expenditure) × 100
where Total capital expenditure = purchases of property and equipment
+ principal payments on finance leases (where separately disclosed)
A reading of 100 means the quarter’s buildout was exactly self-funded. Above 100, operations paid for the capital program and left cash over. Below 100, the shortfall was covered from the balance sheet, from debt, or from equity issuance.
Coverage. CFSR™ is computed only for companies that publish quarterly operating cash flow and capital expenditure in a single primary document. Alphabet and Meta both do, in their quarterly earnings releases. Microsoft and Amazon report on fiscal calendars that do not align to the June quarter in the same disclosure format, so they are excluded rather than estimated. The reading is a two-firm reading and is labelled as such everywhere it appears.
CFSR™ readings, quarter ended June 30
| Company | Operating cash flow (USD M) | Capital expenditure (USD M) | CFSR™ | Source |
|---|---|---|---|---|
| Alphabet, Q2 2025 | 27,747 | 22,446 | 123.6 | Alphabet Q2 2026 earnings release |
| Alphabet, Q2 2026 | 39,069 | 44,924 | 87.0 | Alphabet Q2 2026 earnings release |
| Meta, Q2 2025 | 25,561 | 17,012 | 150.3 | Meta Q2 2026 earnings release |
| Meta, Q2 2026 | 31,862 | 31,078 | 102.5 | Meta Q2 2026 earnings release |
| Combined, Q2 2025 | 53,308 | 39,458 | 135.1 | Axis Intelligence Research calculation |
| Combined, Q2 2026 | 70,931 | 76,002 | 93.3 | Axis Intelligence Research calculation |
Meta’s capital expenditure figure combines purchases of property and equipment ($30,116 million) with principal payments on finance leases ($962 million), matching the basis Meta itself uses to define free cash flow. Alphabet does not separately disclose finance-lease principal in its earnings release cash flow statement, so its capital expenditure line is purchases of property and equipment only.
The arithmetic behind the headline: capital expenditure across the two firms rose 92.6% year over year, while operating cash flow rose 33.1%. Neither business weakened. The gap between those two growth rates is the entire story.
What CFSR™ does not capture. It is a cash measure, not a returns measure. A reading below 100 is not evidence that an investment is bad; a utility building a power plant runs sub-100 for years and earns a perfectly good return. What the reading does establish is the share of the program that depends on external capital, and therefore on financing conditions holding.
Sarah Mitchell: Note what the number is not saying. Alphabet’s operating cash flow grew 41% year over year. That is not a business in trouble. The pressure sits in the denominator: capex doubled to $44.9 billion in one quarter, and the reason is physical. Accelerator generations turn over roughly annually now, so the fleet is being rebuilt while it is being expanded. A ratio driven by a refresh cycle does not need a recession to break below 100 — it just needs the delivery schedule to hold.
Where Is the AI Buildout Money Coming From?
The financing side of the ledger is disclosed with unusual clarity this quarter, because both companies did large raises in the same three months.
| Instrument | Company | Net proceeds (USD M) | Quarter | Source |
|---|---|---|---|---|
| Common stock issuance | Alphabet | 30,499 | Q2 2026 | Alphabet Q2 2026 earnings release |
| Mandatory convertible preferred stock | Alphabet | 19,063 | Q2 2026 | Alphabet Q2 2026 earnings release |
| Debt issuance, net of costs | Alphabet | 24,847 | Q2 2026 | Alphabet Q2 2026 earnings release |
| Long-term debt issuance, net | Meta | 24,910 | Q2 2026 | Meta Q2 2026 earnings release |
| Total external capital raised | — | 99,319 | Q2 2026 | Axis Intelligence Research calculation |
According to Axis Intelligence Research, that $99.3 billion equals 130.7% of the two companies’ combined $76.0 billion of Q2 2026 capital expenditure. Alphabet’s own disclosure is explicit that the June 2026 equity raise — $49.6 billion in aggregate net proceeds across Class A stock, Class C stock, and mandatory convertible preferred — is intended for general corporate purposes “including capital expenditures to scale AI infrastructure and global compute.”
The balance sheet effect is immediate. Alphabet’s long-term debt rose from $46.5 billion at December 31, 2025 to $98.2 billion at June 30, 2026, a 110.9% increase in six months. Meta’s rose from $58.7 billion to $83.7 billion over the same period, up 42.4%. Combined long-term debt for the two firms grew 72.7% in two quarters.
Depreciation is the lagging consequence and it is already visible. The physical side of this spending is tracked separately in our data center construction spending dataset, and the per-megawatt economics in our AI data center cost per MW benchmark. Alphabet’s quarterly depreciation of property and equipment rose 42.1% year over year to $7.10 billion; Meta’s depreciation and amortization rose 46.4% to $6.36 billion. Meta’s operating margin fell to 31% from 43%, and its operating income declined 8% year over year even as revenue grew 28%.
How Exposed Is the Banking System to AI Lending?
This is the question regulators have actually answered, and the answer is more specific than most coverage suggests.
The Federal Reserve Bank of Chicago published a direct measurement in February 2026 using supervisory FR Y-14Q data. Large banks — bank holding companies with $100 billion or more in total consolidated assets — had raised their concentration of commercial and industrial commitments to AI-adjacent industries from about 9% of total commitments (roughly $250 billion) in 2015 to about 13% (roughly $450 billion) in late 2025, of which about $150 billion was drawn.
| Exposure measure | Value | As of | Source |
|---|---|---|---|
| AI-adjacent C&I commitments, large banks | ~$450B | Late 2025 | Federal Reserve Bank of Chicago |
| Of which outstanding | ~$150B | Late 2025 | Federal Reserve Bank of Chicago |
| Share of total C&I commitments | ~13% | Late 2025 | Federal Reserve Bank of Chicago |
| Average outstanding as share of bank total assets | ~0.8% | Late 2025 | Federal Reserve Bank of Chicago |
| Average outstanding as share of tier 1 capital | ~9% | Q3 2025 | Federal Reserve Bank of Chicago |
| Average committed as share of tier 1 capital | ~25% | Q3 2025 | Federal Reserve Bank of Chicago |
| Software C&I commitments | $191B | Late 2025 | Federal Reserve Bank of Chicago |
| Energy + semiconductor C&I commitments | ~$275B | Late 2025 | Federal Reserve Bank of Chicago |
The credit quality signal sits in software, not silicon
The distribution matters more than the total. Around 26% of large-bank commitments to the software industry — roughly $50 billion — carried ratings of B and below as of Q3 2025, against about 13% for the total C&I portfolio. According to Axis Intelligence Research, that is a 2.0× speculative-grade concentration multiple for software relative to the average bank commercial book. The comparable figure for software was about 14% in Q3 2019.
Energy and semiconductors look nothing like this, which is consistent with the supply-side concentration documented in our AI chip market share analysis. Combined B-and-below commitments across those two industries totalled about $15 billion out of roughly $275 billion, and their delinquencies accounted for fewer than 15 basis points of tier 1 capital at every large bank.
The visibility gap
Chicago Fed staff cite a JPMorgan Chase estimate of roughly $1.2 trillion in AI-company-issued debt — more than double large-bank C&I commitments and about eight times the drawn balance. According to Axis Intelligence Research, if that estimate holds, approximately 62.5% of AI-related corporate debt sits outside the large-bank commitment perimeter that supervisory FR Y-14Q data actually observes. The JPMorgan figure is an analyst estimate rather than a regulatory tabulation, and the derived share moves with it: at $900 billion the outside share is 50%; at $1.5 trillion it is 70%.
The Fed’s own Financial Stability Report adds a related detail. Private credit stood at about $1.4 trillion in the second half of 2025, roughly 10% of total U.S. nonfinancial corporate debt, and the software sector had become the largest sector in private credit portfolios. In the first quarter of 2026, accepted redemptions at perpetual business development companies exceeded new inflows for the first time since those vehicles were created.
Sarah Davis: Follow the seniority, not the headline. Banks are not where the AI credit risk concentrates — 0.8% of total assets is a rounding error, and the delinquency data supports the Fed’s own calm reading. The interesting number is the 62.5% that is not in the supervised book. It is in bonds, in private credit, in vehicles whose investors can ask for their money back quarterly. That is not a solvency problem for the banking system. It is a liquidity question about who is holding paper they can’t sell on the day everyone wants to sell it.
Do the AI Adoption Statistics Justify the Spending?
Adoption is the demand-side check on the capex thesis, and the federal statistical system publishes it directly.
The U.S. Census Bureau’s Business Trends and Outlook Survey — a biweekly, nationally representative survey of 1.2 million U.S. businesses — put the national AI use rate at 19.8% as of May 3, 2026, with overall usage moving between 17% and 20% across the December 2025 to May 2026 window. Between 20% and 23% of businesses expected to be using AI within six months.
| Segment | AI use rate | As of | Source |
|---|---|---|---|
| All U.S. businesses | 19.8% | May 3, 2026 | U.S. Census Bureau BTOS |
| Information sector | 39.7% | May 3, 2026 | U.S. Census Bureau BTOS |
| Finance and Insurance | 33.9% | May 3, 2026 | U.S. Census Bureau BTOS |
| Retail Trade | ~14% | May 3, 2026 | U.S. Census Bureau BTOS |
| Firms with 250+ employees | 37% | May 3, 2026 | U.S. Census Bureau BTOS |
| Firms with 100–249 employees | 32% | May 3, 2026 | U.S. Census Bureau BTOS |
| Firms with 1–4 employees | <20% | May 3, 2026 | U.S. Census Bureau BTOS |
Read that table carefully before concluding anything. The 19.8% is firm-weighted, and 95% of U.S. firms have fewer than 50 employees. Weight by employment and the picture inverts. A Federal Reserve FEDS Note published April 3, 2026 compares three surveys and finds the Atlanta Fed’s Survey of Business Uncertainty putting 78% of the U.S. labor force at firms that have adopted AI, and 54% at firms using large language models, from 1,032 responses in November 2025. The Real-Time Population Survey put work-related generative AI use at about 41% of the workforce in November 2025, with daily use at 12%.
So which number is real? All of them, measuring different things. The demand-side counterpart to these adoption rates is capacity, which we track in AI data center market size and AI data center energy consumption statistics. The Fed note is unusually direct about this: the BTOS is the best estimate of the share of U.S. businesses that have adopted AI, the Real-Time Population Survey is the best estimate of the share of the labor force using generative AI at work, and the Survey of Business Uncertainty is a good upper bound on workplace access to AI tools. Anyone citing a single adoption percentage as evidence for or against a bubble is choosing a denominator and hoping nobody checks.
The intensity gap is the number to watch
Daily generative AI use at work stood at 12% in November 2025 against 40.7% any use — under a third. The Fed note names the open question plainly: understanding whether AI infrastructure investment is sustainable depends on the price and quantity of token consumption by end users, and that data does not yet exist in published form.
Sarah Mitchell: Adoption breadth is a poor proxy for inference demand, and both bulls and bears abuse it. Nineteen point eight percent of firms sounds thin until you notice that 56% of U.S. employment sits in the 0.9% of firms with 250 or more employees, and those firms adopt at 37%. And a firm that answers yes may be running one summarisation workflow or fine-tuning against its own corpus — the survey cannot tell those apart, and they differ by orders of magnitude in tokens. The number that would settle the argument is tokens billed, at what price, growing how fast. Nobody publishes it.
Which AI Bubble Warning Signs Do Regulators Actually Cite?
The Federal Reserve Bank of New York surveyed 20 market contacts across broker-dealers, banks, investment funds, and advisory firms during March and April 2026 for the Financial Stability Report’s Survey of Salient Risks. Artificial intelligence ranked third among most-cited potential shocks over the following 12 to 18 months, behind geopolitical risks and an oil shock, and cited by roughly half of respondents against roughly 30% in the fall 2025 survey — where AI had ranked fifth.
The specific AI concerns respondents named were narrower than the general bubble discourse: equity valuations, capital expenditure increasingly funded by debt, and the possibility that widespread AI adoption contributes to labour market weakness. Private credit drew separate concern, partly because AI-driven disruption is affecting the credit quality of some borrowers — the software concentration described above.
| Risk cited | Spring 2026 | Fall 2025 rank | Source |
|---|---|---|---|
| Geopolitical risks | 1st | 2nd | Federal Reserve, May 2026 FSR |
| Oil shock | 2nd | Not in top 12 | Federal Reserve, May 2026 FSR |
| Artificial intelligence | 3rd | 5th | Federal Reserve, May 2026 FSR |
| Private credit | 4th | 9th | Federal Reserve, May 2026 FSR |
| Persistent inflation | 5th | 3rd | Federal Reserve, May 2026 FSR |
Worth stating what the Fed did not say. The May 2026 report does not describe AI as a systemic threat, does not flag bank exposure as a vulnerability, and reports business and household debt-to-GDP falling to levels not seen since the early 2000s. The Chicago Fed’s conclusion was that the fundamentals of the parties receiving AI funding “look appropriate as of now.” Both institutions frame this as tail risk — low probability, correlated consequences — rather than as an unfolding crisis.
Is the AI Bubble the Same as the Dot-Com Bubble?
The comparison is made constantly and it holds in one dimension and fails in another. Our pillar page on AI data center statistics covers the physical buildout; this page covers how it is paid for. The dimension where the dot-com comparison holds is concentration and funding structure: capital expenditure growing far faster than operating cash flow, an increasing share of it debt-financed, and speculative-grade credit clustering in one sector.
The dimension where it fails is the underlying businesses. Alphabet posted 24% revenue growth to $119.8 billion in Q2 2026 with a 34% operating margin, and Google Cloud revenue grew 82% to $24.8 billion. Meta posted 28% revenue growth to $60.8 billion. These are not the profitless companies of 1999. Axis Intelligence Research takes the position that the useful distinction is not “bubble or not” but which layer carries the risk: the operating businesses funding the buildout are demonstrably profitable, while the credit and vehicle layer that finances the rest of the sector — the 62.5% outside the bank perimeter, the private credit funds with quarterly redemption windows — is where the correlated exposure sits.
| Comparison dimension | 2026 evidence | Source |
|---|---|---|
| Operating profitability of largest spenders | Alphabet 34% operating margin, Q2 2026 | Alphabet Q2 2026 earnings release |
| Cash coverage of capex | CFSR™ 93.3, Q2 2026 | Axis Intelligence Research |
| Debt growth at largest spenders | +72.7% combined long-term debt, 6 months | Axis Intelligence Research |
| Bank system exposure | ~0.8% of total assets | Federal Reserve Bank of Chicago |
| Broad equity valuation | Forward P/E well above 16.00 median | Federal Reserve, May 2026 FSR |
Methodology
Collection. Every figure in this dataset was retrieved from a primary document during production between August 3 and August 5, 2026. Company financials come from quarterly earnings releases published by Alphabet Inc. and Meta Platforms, Inc. on their investor relations sites. Regulatory and supervisory figures come from the Board of Governors of the Federal Reserve System and the Federal Reserve Bank of Chicago. Adoption figures come from the U.S. Census Bureau’s Business Trends and Outlook Survey and from Federal Reserve staff analysis of BTOS, the Real-Time Population Survey, and the Atlanta Fed Survey of Business Uncertainty.
CFSR™ computation. Operating cash flow and capital expenditure were taken from the consolidated statements of cash flows in each company’s earnings release. Alphabet: net cash provided by operating activities and purchases of property and equipment. Meta: net cash provided by operating activities, purchases of property and equipment, and principal payments on finance leases. The combined reading sums numerators and denominators before dividing, rather than averaging company ratios, so larger programs carry proportionate weight. All arithmetic was executed in Python and the outputs verified against the source statements line by line.
Basis note. The two companies define capital expenditure differently in their free cash flow reconciliations. Meta includes finance-lease principal; Alphabet’s earnings release does not break it out. CFSR™ uses each company’s own disclosed basis rather than forcing a synthetic common definition, and the asymmetry is disclosed here rather than smoothed away. A finance-lease-exclusive Meta reading for Q2 2026 would be 105.8 rather than 102.5, moving the combined reading to 94.5 rather than 93.3 — a 1.2 point difference that does not change the direction or the sub-100 crossing.
Scope. CFSR™ covers two companies, not the hyperscaler group. It is not a market-wide indicator and should not be cited as one. Microsoft and Amazon are excluded because their disclosure calendars and formats do not permit a like-for-like June-quarter computation from a single primary document.
Methodology edition. August 2026. Any change to the coverage set, the capital expenditure basis, or the aggregation method will be published here with the date of the change, and prior readings will not be restated silently.
Known constraints. The $1.2 trillion AI-issued debt figure underlying the 62.5% visibility gap is an analyst estimate cited by Federal Reserve Bank of Chicago staff, not a regulatory tabulation; the sensitivity range is given inline. Survey-based adoption rates carry sampling and question-framing differences that the Federal Reserve FEDS Note discusses at length. Federal Reserve survey-of-risks percentages are read from a published figure and are reported here as ranks plus approximate shares.
About This Dataset
Contents. 123 observations covering hyperscaler quarterly cash flow and capital structure, U.S. bank AI-adjacent credit exposure, U.S. business AI adoption by sector and firm size, broad equity valuation measures, and Axis Intelligence Research derived metrics including CFSR™.
Temporal coverage. 2015 to June 30, 2026, with the bulk of observations dated between Q3 2025 and Q2 2026.
Spatial coverage. United States, with company financials reported on a global consolidated basis.
License. CC BY 4.0. Free to use, share, and adapt with attribution.
Citation line. Axis Intelligence Research, AI Bubble Statistics 2026, 2026.
Distribution: Hugging Face · Kaggle · GitHub
Cite This Research
APA Axis Intelligence Research, Mitchell, S., & Davis, S. (2026, August 5). AI bubble statistics 2026: Capex, credit, and adoption data. Axis Intelligence. https://axis-intelligence.com/ai-bubble-statistics/
MLA Axis Intelligence Research, Sarah Mitchell, and Sarah Davis. “AI Bubble Statistics 2026: Capex, Credit, and Adoption Data.” Axis Intelligence, 5 Aug. 2026, axis-intelligence.com/ai-bubble-statistics/.
Chicago Axis Intelligence Research, Sarah Mitchell, and Sarah Davis. “AI Bubble Statistics 2026: Capex, Credit, and Adoption Data.” Axis Intelligence, August 5, 2026. https://axis-intelligence.com/ai-bubble-statistics/.
Frequently Asked Questions
What is the CFSR™ and how is it calculated?
The Capex Funding Self-Sufficiency Ratio is an Axis Intelligence Research metric equal to net cash provided by operating activities divided by capital expenditure, times 100. The Q2 2026 combined reading for Alphabet and Meta is 93.3, computed from $70.93 billion of operating cash flow against $76.00 billion of capital expenditure. Both inputs come from the companies’ quarterly earnings releases.
Did Alphabet really report negative free cash flow?
Yes. Alphabet’s Q2 2026 earnings release reports free cash flow of negative $5.855 billion for the quarter ended June 30, 2026, from $39.07 billion of operating cash flow less $44.92 billion of capital expenditure. Trailing twelve month free cash flow remained positive at $53.27 billion.
How much AI-related debt do U.S. banks actually hold?
Large U.S. banks held roughly $450 billion in committed and $150 billion in outstanding C&I exposure to AI-adjacent industries in late 2025, per Federal Reserve Bank of Chicago analysis of FR Y-14Q supervisory data. Outstanding exposure averaged about 0.8% of bank total assets and about 9% of tier 1 capital, with committed exposure closer to 25% of tier 1 capital.
Which AI-adjacent sector carries the weakest credit quality?
Software. About 26% of large-bank commitments to software companies carried B-and-below ratings as of Q3 2025, against about 13% for the overall C&I portfolio and about 5% for energy and semiconductors. Software commitments grew from $150 billion in early 2022 to $191 billion in late 2025.
What share of U.S. businesses use AI in 2026?
19.8% as of May 3, 2026, on a firm-weighted basis, per the U.S. Census Bureau’s Business Trends and Outlook Survey. Employment-weighted estimates are far higher: the Atlanta Fed’s Survey of Business Uncertainty put 78% of the labour force at firms that have adopted AI as of November 2025.
Is the Federal Reserve calling AI a bubble?
No. The May 2026 Financial Stability Report describes asset valuation pressures as elevated and lists AI among the risks cited by surveyed market contacts, but does not characterise AI as a bubble or flag it as a systemic vulnerability. Chicago Fed staff framed AI lending as tail risk while describing current borrower fundamentals as appropriate.
How much did hyperscaler capex grow year over year?
Alphabet’s Q2 2026 capital expenditure of $44.92 billion was double the $22.45 billion of Q2 2025. Meta’s rose to $31.08 billion from $17.01 billion. Combined, the two grew 92.6% year over year while their combined operating cash flow grew 33.1%.
What would reverse the CFSR™ reading?
Three things, in order of likelihood: capex guidance coming down, cloud backlog converting to billed revenue faster than infrastructure is delivered, or depreciation schedules lengthening. Alphabet guided 2026 capital expenditure to $195–205 billion and Meta to $130–145 billion, so the first is not in the current plan.
Why are Microsoft and Amazon excluded from the CFSR™?
Their fiscal calendars and cash flow disclosure formats do not permit a like-for-like June-quarter computation from a single primary document. Estimating their inputs would have produced a more impressive-looking index built partly on inference, which is the opposite of what the metric is for.
How does 2026 compare with the dot-com peak on valuation?
The Federal Reserve’s May 2026 report places the S&P 500 forward price-to-earnings ratio well above its 16.00 historical median and the equity premium near a twenty-year low, but does not publish a direct dot-com comparison. Axis Intelligence Research does not restate third-party peak-year multiples that it has not verified against a primary source.
Corrections and data queries: [email protected] ·
Related Axis Intelligence Research: AI Data Center Market Size · AI Data Center Statistics · Data Center Construction Spending · AI Chip Market Share · AI Data Center Energy Consumption Statistics AI Capex Tracker