Intel — Sustainability Impact Analysis
Used SQL to analyze 601,740 repurposed devices from Intel's 2024 sustainability program, looking for where the program was underperforming environmentally and building a data-driven recommendation from what the data showed.
The Task
Intel's repurposing program extends the life of existing laptops and desktops rather than manufacturing new ones, which reduces e-waste, energy consumption, and CO₂ emissions. The question I was asked to answer was whether the program was targeting the right devices in the right regions to get the most environmental impact out of each one.
I worked with two datasets, device data and impact data, and was asked to join, clean, bucket, and analyze them to surface patterns and produce a concrete recommendation. Everything was written in SQL.
The Approach
The analysis ran in four stages. I started by joining the two tables on device_id
and calculating a device_age column from the model year. From there I used a
CASE WHEN clause to group devices into newer (0 to 3 years), mid-age (4 to 6 years),
and older (7 or more years) buckets. I then aggregated by device type, age bucket, and region
to find where the energy savings and CO₂ reductions were actually concentrated. The final
query used multiple CTEs to calculate each device type's percentage
contribution to regional totals.
The core bucketing query that most of the analysis built on:
SELECT d.*, i.*, (2024 - d.model_year) AS device_age, CASE WHEN (2024 - d.model_year) <= 3 THEN 'newer' WHEN (2024 - d.model_year) <= 6 THEN 'mid-age' WHEN (2024 - d.model_year) > 6 THEN 'older' END AS device_age_bucket FROM intel.device_data d INNER JOIN intel.impact_data i ON d.device_id = i.device_id
Key Findings
Volume vs. per-unit impact
The program repurposes mostly newer devices (317,191) but older devices save 48 kWh per unit compared to 19 kWh for newer ones. The program is getting more volume but leaving per-unit environmental return on the table.
Laptops drive total savings
Laptops make up 408,064 of the 601,740 devices repurposed, more than twice as many as desktops. Average savings per device are nearly the same for both, so laptop volume is what drives the total CO₂ numbers.
Region changes the CO₂ math
Energy savings per device are roughly equal across all three regions at around 25.7 kWh, but Asia saves 0.0155 tons of CO₂ per device compared to 0.0064 tons in Europe. Asia's power grid relies heavily on coal, so each kWh saved there prevents significantly more emissions.
Scale of the program
601,740 devices at around 25.7 kWh each adds up to roughly 15.5 million kWh saved and 6,678 tons of CO₂ in a single year, which is about the equivalent of taking 1,500 cars off the road.
The Recommendation
Route older laptops to Asia.
Older devices save 48 kWh per unit compared to 19 kWh for newer ones, and laptops already make up the majority of what gets repurposed. Asia's coal-heavy grid means each device saves 0.0155 tons of CO₂, more than double the 0.0064 tons saved in Europe. Combining the highest per-unit energy savings with the highest-impact region produces the best possible environmental return per device.
The main constraint is supply. Only 20,239 older devices were repurposed in 2024, so Intel would also need to expand collection partnerships to source more of them. Transportation costs to Asia could also offset some of the gains, but that analysis would require cost data that was not available in this dataset.
What I Took Away
The SQL itself was not the hard part. The interesting moment was when two separate findings connected. I had older devices saving more energy per unit in one query, and Asia preventing more CO₂ per kWh in another. When I looked into why Asia's numbers were so different, the answer was coal. That is when the recommendation became obvious.
It is a good example of why data analysis is not just about writing correct queries. It is about knowing when two results are actually pointing at the same thing.