DEVP 226 · Final Project · UC Berkeley

NEVI, Utilities, and Installers Bottlenecks in U.S. EV Charging Deployment

An empirical analysis of 9,229 DCFC stations deployed across all 50 states (2020–2024), quantifying how federal policy, state institutions, and supply chains shape America's transition to electric mobility.

SK
Sotaire Kwizera
Data Analyst · Graduate Student at UC Berkeley
Scroll to explore
+153.5%
National Deployment
Acceleration (Post-NEVI)
9,229
DCFC Stations Analyzed
(2020–2024)
50+1
States Covered
(All states + D.C.)
4
Targeted Policy
Recommendations
Key Findings

Three core insights from the data

Analyzing five years of charging infrastructure deployment reveals that institutional capacity—not hardware or capital—determines whether federal funding translates into operational stations.

01

Federal policy works, but state capacity is the multiplier.

The Infrastructure Investment and Jobs Act produced a 153.5% national acceleration in DCFC deployment—from 92.1 to 233.4 stations/month. Yet state-level results vary enormously.

  • Texas+520.8%
  • California+8.7%
  • Mississippi+860%*

*Off a near-zero baseline (0.21 → 2.00 stations/month).

02

Institutional bottlenecks dominate deployment time.

Permitting (Stage C) and utility interconnection (Stage E) account for 65–85% of total deployment lag—not hardware manufacturing. Reducing institutional friction yields immediate gains without waiting for new technology.

Interconnection 40–50%
Permitting 25–35%
Hardware 20–25%
03

Network structure mediates a scale-vs-quality tradeoff.

ChargePoint's capital-light model deploys 52.17 stations/month with 85–90% uptime, while Tesla's vertical integration yields fewer stations (29.65/month) at >99% uptime—a real-time test of coordination economics.

  • ChargePoint share33.9%
  • Tesla share19.3%
  • Networks tracked93
Interactive Tool

Explore the data yourself

A live Streamlit dashboard lets you filter 80,000+ U.S. charging stations by state, network, charger type, and year—then ask EVA, an embedded AI assistant powered by Google Gemini, natural-language questions about the data.

⚡ ev-charging-dashboard.streamlit.app
🤖
EVA Ask me anything
Total80,597
DCFC15,056
States51
Networks93
Interactive U.S. Map
🗺️

Interactive Map

Filter 80,000+ stations by state, network, charger type, and year

🤖

EVA AI Assistant

Ask plain-English questions; Google Gemini answers with live data

📊

State Comparison

Per-capita and density metrics, choropleth maps, network breakdowns

📥

Export Data

Download filtered station data as CSV for your own analysis

Research Paper

The full empirical study

A peer-reviewed-style research paper presenting the methodology, supply chain framework, quantitative analysis, and policy recommendations. Built on the Alternative Fuels Data Center (AFDC) database via the NREL API.

  • I.Introduction
  • II.Background & Literature Review
  • III.Conceptual Framework
  • IV.Methodology
  • V.Findings: Bottleneck Analysis
  • VII.Quantitative Analysis
  • VIII.Policy Implications
  • IX.Conclusion
NEVI, Utilities, and Installers
Abstract. This paper analyzes deployment bottlenecks in U.S. DC fast-charging infrastructure using a 9,229-station sample (2020–2024). A pre/post-NEVI deployment-rate comparison documents a 153.5% national acceleration, while state-level variation reveals that institutional capacity—not hardware—determines whether federal funding translates into operational stations…
How it was built

Methodology & Tech Stack

1

Data Acquisition

Pulled 80,597 charging stations from NREL's Alternative Fuels Data Center API and validated 9,229 DCFC stations with operational dates (2020–2024).

Python NREL API requests
2

Velocity Analysis

Applied a temporal-comparison methodology: pre-NEVI (2020–2021) vs. post-NEVI (2023–2024) deployment rates by state and network—24 vs. 24 months.

pandas numpy
3

Visualization

Built an interactive Streamlit dashboard with Plotly maps, choropleths, and time series. Custom CSS for a polished UI and floating EVA chat.

Streamlit Plotly Folium
4

AI Layer

Integrated Google Gemini (gemini-2.0-flash) to answer natural-language questions, grounded in the live filtered dataset's summary statistics.

google-genai Gemini 2.0

Let's connect.

Interested in EV infrastructure, energy economics, or data-driven policy research? I'd love to chat.