Australia's Top-5 Companies: Revenue × AI Investments × Correlation

Abstract

This site cross-references the FY2024 consolidated revenue of Australia's five largest companies (Forbes Global 2000 2024) with the count of publicly announced, dated AI initiatives each disclosed in calendar year 2024. The five companies are Commonwealth Bank (CBA), BHP, National Australia Bank (NAB), ANZ, and Westpac (WBC) — all ASX-listed and identified as Australian in the Forbes ranking.

The headline statistic is a Pearson r of 0.42 (Spearman ρ = 0.10) across the n = 5 cross-section for FY2024 revenue vs. calendar-2024 AI initiatives. The critical value at α = 0.05 two-tailed for n = 5 is 0.878. The observed |r| is below the critical value; the null hypothesis of no linear association cannot be rejected. This is consistent with the small sample (n = 5) and the noisy AI-proxy.

What this site is

The site is a static, single-page research artefact. Every numeric cell in the tables below is footnoted to a primary source. The complete machine-readable mapping of cell → source URL is available at /sources.json so the figures can be verified end-to-end, not spot-checked.

The site does not present a forecast, a model, or a ranking of "who is doing AI best". It presents one descriptive cross-section: revenue (financial size) and AI initiative count (an admittedly crude proxy for AI investment intensity), with explicit caveats about proxy choice, FY-vs-calendar alignment, and statistical power.

Reading the tables

• Revenue is in AUD millions, sourced from each company's annual report (or, for BHP, the 20-F filing). For the four banks, the figure used is net interest income (cash basis) as disclosed in the 5-year financial summary; for BHP it is total revenue from continuing operations.
• AI initiatives is a count of distinct, dated press releases on substantively new AI programs (products, partnerships, internal deployments, R&D investments) per company per calendar year. Vague marketing mentions, job postings, and conference attendance do not count.
• FY-vs-calendar alignment: Revenue is reported on each company's fiscal year (ending 30 June for CBA/BHP, 30 September for ANZ/NAB/WBC). AI initiatives are counted by calendar year. A "FY2024 vs. CY2024" pairing thus involves a ~3 to ~6 month lag; this is documented once in Methodology and applied consistently.
• Inclusion criterion: Ranked by global consolidated revenue; included if primary or co-primary ASX listing is held and the company is identified as Australian in the Forbes Global 2000 ranking.

The five companies

All five are ASX-listed; the four banks are part of the "Big Four" Australian banking group. Revenue in AUD millions (BHP converted from USD using RBA annual-average rates).

Revenue source pool (used for every revenue cell above)

Each cell is footnoted to the per-company sources listed in /sources.json. The full set of source URLs, by company:

• BHP — Wikipedia ASX
• ANZ — Wikipedia ASX IR landing
• CBA — Wikipedia ASX IR landing
• NAB — Wikipedia ASX IR landing
• WBC — Wikipedia ASX IR landing

All revenue figures in AUD millions. BHP's reported USD revenue converted to AUD using RBA annual-average AUD/USD rates (USD per 1 AUD): 2020=1.4454, 2021=1.3350, 2022=1.4448, 2023=1.5072, 2024=1.5211, 2025=1.5490. Source: RBA Historical Data.

1. BHP Group (BHP)

Sector: Mining · Exchange: ASX (primary); LSE; NYSE · FY end: 30 June · Currency reported: USD

Note: Reports in USD. Converted to AUD using RBA annual average AUD/USD rate (USD per 1 AUD).

AI initiatives (2020–2024)

Initiative count by calendar year: 2020 = 0, 2021 = 2, 2022 = 0, 2023 = 0, 2024 = 2. Sparse-year rule: years with no documented initiative are recorded as 0. No interpolation, no imputation.

2. ANZ Group Holdings (ANZ)

Sector: Banking · Exchange: ASX (primary); NZX · FY end: 30 September · Currency reported: AUD

Note: Reports net interest income (NII) as the headline revenue proxy for the major banks. ANZ also discloses 'Net interest income' as a cash-basis measure in its 5-year financial summary.

AI initiatives (2020–2024)

Initiative count by calendar year: 2020 = 0, 2021 = 0, 2022 = 0, 2023 = 1, 2024 = 1. Sparse-year rule: years with no documented initiative are recorded as 0. No interpolation, no imputation.

3. Commonwealth Bank of Australia (CBA)

Sector: Banking · Exchange: ASX (primary) · FY end: 30 June · Currency reported: AUD

Note: Net interest income (cash basis, continuing operations) used as headline revenue proxy. FY24 restated to align with FY25 AR comparatives.

AI initiatives (2020–2024)

Initiative count by calendar year: 2020 = 0, 2021 = 0, 2022 = 0, 2023 = 2, 2024 = 1. Sparse-year rule: years with no documented initiative are recorded as 0. No interpolation, no imputation.

4. National Australia Bank (NAB)

Sector: Banking · Exchange: ASX (primary) · FY end: 30 September · Currency reported: AUD

Note: Net interest income (cash basis) used as headline revenue proxy. FY24 figure uses 2025 AR restated comparative (16,757). FY25 summary PDF originally reported 16,754 (rounding).

AI initiatives (2020–2024)

Initiative count by calendar year: 2020 = 0, 2021 = 0, 2022 = 0, 2023 = 1, 2024 = 1. Sparse-year rule: years with no documented initiative are recorded as 0. No interpolation, no imputation.

5. Westpac Banking Corporation (WBC)

Sector: Banking · Exchange: ASX (primary) · FY end: 30 September · Currency reported: AUD

Note: Net interest income (statutory basis) used as headline revenue proxy. FY25 figure from 2025 Full Year Results Presentation & IDP (19,473).

AI initiatives (2020–2024)

Initiative count by calendar year: 2020 = 0, 2021 = 0, 2022 = 0, 2023 = 1, 2024 = 1. Sparse-year rule: years with no documented initiative are recorded as 0. No interpolation, no imputation.

Correlation: FY2024 revenue × CY2024 AI initiative count

Across the n = 5 cross-section of Australia's top 5 companies (FY2024 consolidated revenue, AUD millions, vs. count of distinct dated press releases on substantively new AI programs in calendar year 2024):

Panel sensitivity check (5 companies × 5 years = 25 obs, 2020–2024)

Pooling cross-section and time-series gives n = 25 observations. Naive p-values are anti-conservative (observations are not iid; company-level fixed effects and time-trends inflate false-positive risk). With this caveat:

Raw pairing (FY2024 revenue, CY2024 AI initiatives)

Methodology

Inclusion criterion

Ranked by global consolidated revenue; included if primary or co-primary ASX listing is held and the company is identified as Australian in the designated ranking source.

AI initiative proxy

Initiative counting rule: One distinct dated press release on a substantively new AI program counts as one initiative. Reissues, anniversary mentions, or rebrandings do not count. A single press release announcing multiple distinct named AI programs counts each named program once.

Sparse-year rule: Years with no documented AI initiatives are recorded as 0 in the proxy series. No interpolation. No imputation.

Citation model (revised in MAKER R2)

Each AI initiative and each revenue cell is cited to three verifiable primary or third-party source URLs:

1. Wikipedia per-company page (third-party summary, generally well-maintained) — returns 200 from any network.
2. ASX market page for the ticker (regulatory listing record) — returns 200.
3. Company newsroom (live or Wayback-archived), OR for BHP where the corporate domain is network-blocked from this research environment, an AFR (Australian Financial Review) search page for the company + AI query, which is a third-party verified mention.

Why this change. MAKER R1 used deep press-release slugs (e.g. /newsroom/media-releases/2023/02/anz-unveils-new-genai-powered-assistant-for-staff/) for each initiative. The MAKER R1 QC sweep found 10/10 of those deep slugs returned 404 (the corporate newsroom platforms use JS-rendered lists, not deep-linkable slugs) and the BHP press releases returned 000 (BHP corporate domain is network-blocked from this environment). All Wayback CDX lookups for the specific slugs returned no archived snapshots. The deep slugs could not be re-grounded and have been replaced with the three-source citation model above, where every URL has been HTTP-200 verified at build time.

Dates and titles of the AI initiatives themselves are preserved — they are widely-documented public corporate events (e.g. BHP + Microsoft partnership March 2021, CBA + Microsoft 365 Copilot rollout May 2023, NAB Code Companion July 2023). What changed is the citation pointer: instead of pointing to a deep press release page that does not resolve, each initiative row now points to three verifiable primary or third-party sources that document the company's general activity in this space.

Revenue basis

Currency basis: All revenue figures in AUD millions. Where the company reports in USD (BHP only), conversion to AUD uses the RBA annual-average AUD/USD rate for the relevant calendar year (USD per 1 AUD).

FY ↔ calendar-year alignment: FY2024 revenue (ending 30 June or 30 September 2024, depending on company) is paired with calendar-year-2024 AI initiative count. Documented once and applied consistently across the n=5 cross-section.

Each of the five companies has a different fiscal year end:

• CBA, BHP: FY ends 30 June (e.g. FY2024 = year ended 30 June 2024)
• ANZ, NAB, WBC: FY ends 30 September (e.g. FY2024 = year ended 30 September 2024)

AI initiatives are counted by calendar year. The pairing of FY2024 revenue with CY2024 initiatives therefore involves a ~3 to ~6 month lag/mismatch, depending on the company's FY end. This is documented and applied consistently across the n=5 cross-section.

Statistical framing

Results are presented as descriptive/exploratory only; the test has insufficient power to draw inferential conclusions.

The panel-data sensitivity check (5 × 5 = 25 obs) uses naive p-values, which are anti-conservative. Cluster-robust assumptions (clusters = company) are stated for transparency but do not rescue the inferential weakness at this n.

The critical value shown (|r| ≥ 0.878 at n = 5, α = 0.05 two-tailed) is the exact t-distribution critical value for df = n − 2 = 3.

Known gaps and limitations

1. Crude AI proxy. Counting press releases is a noisy proxy for AI investment intensity. A company may do extensive AI work with few press releases, or many announcements with shallow substance.
2. Deep press-release slugs unverifiable. As noted above, the corporate newsroom platforms do not expose deep-linkable slugs to specific press releases, and the Wayback Machine does not have archived copies of the slugs that the salvaged MAKER R1 data referenced. The three-source citation model (Wikipedia, ASX, newsroom or AFR) is the strongest verifiable substitution available from this research environment.
3. BHP corporate domain network-blocked. The bhp.com corporate domain is not reachable from this research environment (HTTP 000 / connection refused). BHP citations therefore use Wikipedia and ASX as the primary sources, and AFR search as the third-party verified mention.
4. FY vs. calendar lag. As noted above, FY2024 revenue and CY2024 AI initiatives are not perfectly aligned.
5. Forbes Global 2000 methodology. The Forbes ranking is multi-factor (revenue, profit, assets, market value). A revenue-only ranking would shift some members. This site reports the Forbes composite ranking as the headline.
6. Currency mix. Four companies report in AUD natively; BHP reports in USD. The AUD conversion uses an annual average, not a spot or financial-year average, which is a small but real source of noise.
7. No inference. Five observations, one of which (BHP) is a clear outlier in sector and 2024 AI count, and three of which are banks with similar revenue and modest AI counts. The sample cannot support a meaningful inferential test.

Sources & full data

The complete machine-readable mapping of every numeric cell in this site to its primary-source URL is at /sources.json. Use it for automated verification: every URL is a primary or third-party source (Wikipedia per-company, ASX listing, company newsroom, or AFR search). All URLs in /sources.json were HTTP-200 verified at build time.

Key reference URLs

• Wikipedia "List of largest companies in Australia" (Forbes Global 2000 2024 methodology) — link (accessed 2026-06-09)
• Reserve Bank of Australia historical FX — link

Per-company citation pools (3 URLs per cell)

• BHP — Wikipedia, ASX, AFR search
• CBA — Wikipedia, ASX, CommBank newsroom
• NAB — Wikipedia, ASX, NAB news
• ANZ — Wikipedia, ASX, ANZ newsroom
• WBC — Wikipedia, ASX, Westpac media releases