Case Study: Strategic Revenue Forecasting for Alphabet (2025–2028)

Executive Summary

This project aims to predict Alphabet (Google) revenue trajectory through 2028. By moving beyond simple linear regression to a segment-based analysis and leveraging Meta’s Prophet model with external regressors. This was my first time using Prophet model and I was testing how to construct a dynamic forecasting engine.

The model projects Alphabet’s revenue to reach approximately $560 Billion by 2028, driven not by its traditional search monopoly, but by a fundamental shift to Cloud infrastructure.

Segment Analysis - The Hypothesis :

A core thesis based on market analysis (post-Q3 2025). I performed a segment analysis because looking at total Revenue do not tell the correct picture. My approach was the give conservative growth estimate for each segment and visualize how Google revenue will develop in three years.

The Thesis:

Historically, Alphabet has been an advertising monopoly. My hypothesis is that Google Cloud is growing faster than rest of the business and that this shift will fundamentally reshape the company’s revenue structure. Cloud is positioned to become the primary driver of revenue growth.

Revenue growth estimates used in the analysis.

1. Maturation of Ads (The 5% Growth): Search revenue is reaching saturation. I have used estimate that Ad revenue growth tapering from 8% (2026) down to 5% (2028). It is the cash cow funding the AI transition, but it is no longer the primary growth engine.
   

2. The AI Infrastructure Supercycle: We are entering a major inference phase. Companies that spent 2024/2025 training models must now run them at scale. This operational demand flows directly to Google Cloud Platform (GCP).
   

3. The Growth Delta: While Ads grow linearly, Cloud is expected to grow exponentially sustaining >20% compound growth through 2028.

Forecast: Using Meta´s Prophet

To model SaaS company business behavior, I found out the Prophet model which should be quite good to estimate future Revenues. This model should be more versatile than basic linear regression models.

What is Prophet?

Prophet is an open-source forecasting procedure developed by Facebook's Data Science team in 2017. It is designed for business time series that have strong seasonal effects and historical data shifts.

Why is this the proper model for Revenue in the Alphabet case?

Traditional financial models often fail on tech revenue for three reasons. Prophet solves them:

Modeling the Seasonality: Alphabet’s business is cyclical. Q4 is always ~15% higher than Q3 due to ad spending. Linear models treat this as noise. Prophet natively decomposes the data using a fourier series to approximate this complex seasonality ensuring we don't mistake a Q1 drop for a business failure.

Regressors: No company exists in a vacuum. I chose Amazon’s revenue as a regressors has a Pearson Correlation of 0.9925 with Alphabet. By adding Amazon’s growth as a "signal" (regressor) into the model, this allowed Alphabet’s forecast to "ride the wave" of the broader Cloud/E-commerce market.

Changepoint Detection: Financial history is not smooth; it has changes (e.g., the 2023 AI Boom). Prophet automatically detects these changes as a moments where the trend line bends and adjusts the future trajectory accordingly without manual intervention.

Technical: The Science & The Code

Prophet is based on a Generalized Additive Model (GAM). Unlike "black box" Neural Networks, GAMs are mathematically interpretable. Prophet is specifically designed for business time series, where trends, seasonality and irregular events all need to be modeled in a transparent and modular way.

The core equation driving our forecast is:

y(t) = g(t) + s(t) + β·X(t) + ε(t)

• g(t) - Trend: Long-term, non-periodic growth (e.g., Cloud adoption).

• s(t) - Seasonality: Recurring patterns such as quarterly earnings cycle.

• β·X(t) - The Regressor: External factors added to model to improve the forecast e.g. "Amazon Factor" (Market momentum).

• ε(t) - Error term: Unexplained variation or random noise.

  
  

Scenario Analysis: The Concept

Instead of a single prediction, I used three scenarios for future, based on the Cloud supercycle hypothesis. The scenarios apply growth spreads of 5%, 10%, and 15% to create a clear and actionable risk profile.

Bear Case (5% Sector Growth):

• Simulates a recession or regulatory breakup.

• Strategic Value: Answers "Is Alphabet resilient if the market crashes?

Base Case (10% Sector Growth):

• The Consensus: Aligns with current Wall Street forecasts for Amazon/Tech.

• Result: Alphabet reaches ~$560 Billion by 2028.

Bull Case (15% Sector Growth):

• Assumes Generative AI adoption accelerates Cloud spend faster than retail.

• Result: Alphabet breaks $600 Billion revenue.

Future Roadmap: Improving the Prediction

To evolve this from a prototype to a production-grade financial engine, I would recommend:

Macro-Economic Regressors: Tech revenue is highly sensitive to the cost of capital. Adding the Fed Funds Rate or GDP Growth as secondary regressors would help the model predict downturns that Amazon alone might miss.

Segmented Forecasting: Instead of predicting "Total Revenue," Data should be split into segments Google Cloud vs. Google Services as I did in the segment analysis.

Cloud grows at ~30% while Ads grow at ~5%. Blending them hides the true drivers. Running two separate Prophet models and summing the results could be more accurate.

Attachments: