The AI infrastructure cycle created the most polarized big-tech earnings divergence in a decade. Two years of data tells the whole story.
Nvidia printed a 17x EPS increase across two fiscal years. Meta's efficiency reset nearly tripled earnings. Amazon turned a year of net losses into $5.53 EPS. Meanwhile Apple ground out 5% total EPS growth across the same stretch. This is not a sector story — it's a company-by-company story about positioning versus the AI capex cycle.
AAPL, MSFT, GOOGL, AMZN, META, NVDA: EPS by Year
Note: Nvidia's fiscal year ends in January. FY2023 = Feb 2022–Jan 2023; FY2024 = Feb 2023–Jan 2024; FY2025 = Feb 2024–Jan 2025. All others are calendar year.
| Company | EPS 2022 | EPS 2023 | EPS 2024 | 2-Yr Change |
|---|---|---|---|---|
| NVDANvidia | $1.74* | $12.96* | $29.76* | +17.1x |
| METAMeta | $8.59 | $14.87 | $23.86 | +2.8x |
| AMZNAmazon | –$0.27 | $2.90 | $5.53 | Flipped |
| GOOGLAlphabet | $4.56 | $5.80 | $8.04 | +76% |
| MSFTMicrosoft | $9.65 | $11.45 | $12.41 | +29% |
| AAPLApple | $6.11 | $6.13 | $6.42 | +5% |
*Nvidia fiscal year. FY2023 ends Jan 2023; FY2024 ends Jan 2024; FY2025 ends Jan 2025. Sources: SEC 10-K filings, Bloomberg.
Revenue Growth: Who Is Actually Getting Bigger
| Company | Revenue 2022 | Revenue 2024 | 2-Yr Growth |
|---|---|---|---|
| NVDA | $26.9B* | $130.5B* | +385% |
| AMZN | $514B | $638B | +24% |
| MSFT | $198B | $245B | +24% |
| GOOGL | $282B | $350B | +24% |
| META | $116B | $164B | +41% |
| AAPL | $394B | $391B | –1% |
*Nvidia FY2023 (ends Jan 2023) and FY2025 (ends Jan 2025). Apple revenue declined slightly due to China headwinds and iPhone mix.
What Drove the Divergence
Nvidia
AI GPU monopoly pricing. H100 and H200 ASPs of $25K–$40K per unit with 60%+ gross margins. Every hyperscaler and frontier AI lab is a captive buyer.
Meta
Year of Efficiency: headcount cut from 87K to 67K in 2023, then held flat while revenue accelerated. AI-driven ad targeting lifted CPMs. Operating margins expanded from 20% to 41%.
Amazon
AWS margin expansion drove the swing from loss to $5.53 EPS. AWS operating income grew from $22.8B in 2022 to $39.8B in 2024. Retail segment restructuring reduced fulfillment costs by $5B+.
Microsoft
Azure AI revenue integration (Copilot, OpenAI partnership) added ~$4B in incremental revenue by 2024. Steady compounding on cloud re-signed contracts. EPS grew reliably but without a step-change.
Alphabet
2022 was an advertising recession year. The 2023–2024 recovery combined with YouTube growth and Google Cloud crossing $11B quarterly revenue drove EPS nearly doubling from the 2022 trough.
Apple
China revenue headwinds, iPhone ASP plateauing, and lack of a generative AI hardware catalyst kept growth muted. Services grew to 24% of revenue but couldn't offset hardware stagnation.
Margins Tell the Real Story
EPS can be gamed with buybacks. Operating margin expansion is harder to fake — it reflects real structural improvement.
Efficiency reset + AI ad targeting
AI GPU premium pricing power
AWS mix shift + fulfillment restructuring
Steady; Copilot uplift still building
Ad market recovery + cloud scale benefits
Services mix improving but hardware drag remains
What This Means for Investors and Founders
Three things I take from this data as a VC who watches these numbers closely:
Nvidia is a capex proxy, not a product company
NVDA's earnings are not sustainable at the 2024 growth rate. They are a direct readthrough for AI infrastructure spend. When hyperscalers pull back capex — and they will, eventually — Nvidia's EPS will compress faster than its revenue.
Meta is the best-managed big-tech stock of the cycle
Going from 87K employees to 67K while nearly tripling EPS is a masterclass in corporate efficiency. Most companies that cut headcount 23% see culture destruction and revenue miss. Meta tightened and accelerated. That's rare.
Apple's flat EPS is a warning signal for hardware-first models
When you have 98% brand loyalty and still can't grow EPS by more than 5% over two years, the hardware plateau is real. Services are good but not enough. The AI hardware supercycle will be crucial — if Apple misses it, the flat line gets worse.
The EPS divergence between these six companies is not noise.
It is a signal that AI infrastructure positioning — not legacy moats or brand strength — is now the primary driver of earnings growth for the next decade.
Track real-time big tech earnings multiples on the Big Tech Earnings Dashboard.
Data sourced from SEC 10-K filings and Bloomberg consensus. Nvidia fiscal years noted separately. Originally published in the Trace Cohen newsletter.