The customer: navigating a once-in-a-generation industry shift
Our customer, the global power technology leader, has long been a leader in powering Class 7-8 heavy-duty trucks. But as the transportation industry undergoes a rapid transition toward alternative powertrains, electric, compressed natural gas (CNG), hybrid, and beyond, the company faces a fundamental strategic question:
Which technologies will win? Where will they scale? When will they become viable?
The challenge wasn’t just forecasting vehicle adoption. It was about understanding whether the energy infrastructure itself could support that transition.
Existing tools and datasets provided fragmented answers. There was no unified way to:
- Predict adoption trends across states and regions
- Validate whether fueling and charging infrastructure could sustain those trends
- Identify where infrastructure gaps could become strategic investment opportunities
What the customer needed was not just a forecast but a defensible, system-level view of the future energy ecosystem for which they partnered with Onix.
The challenge: when forecasting alone isn’t enough
As internal planning teams explored future scenarios, a critical gap became clear: Traditional forecasting models could project adoption curves, but they ignored real-world constraints.
- Fleet operators won’t adopt vehicles they can’t fuel.
- Utilities can’t scale instantly to meet demand spikes.
- Infrastructure buildout doesn’t happen uniformly across regions.
Without factoring in these realities, projections risked being theoretically accurate but practically unusable. The global power technology leader with Onix was set out to test a new concept: Could a unified analytics platform connect vehicle demand, infrastructure readiness, and energy capacity into a single predictive framework?
The solution: building a predictive energy intelligence platform
To answer this, the Onix team developed a three-layered predictive analytics platform, bringing together adoption modeling, infrastructure mapping, and energy capacity validation into a single system.
1. Vehicle adoption forecasting
Using historical vehicle registration data from S&P Global, advanced time-series models were built with Prophet to project adoption trends for electric and CNG trucks across all U.S. states through 2040.
- State-level forecasts by region and timeframe
- Comparative modeling approaches, including evaluation of the Bass Diffusion Model
- Long-term projections grounded in real-world historical signals
While multiple modeling techniques were explored, the final approach prioritized accuracy, interpretability, and scalability.
2. Infrastructure mapping
Forecasting demand was only half the equation. The platform mapped critical infrastructure layers:
- CNG fueling station availability
- EV charging networks
- Energy production and distribution nodes
This ensured that every adoption curve could be evaluated against a simple constraint: can this transition actually happen on the ground? By linking vehicle demand to infrastructure availability, the platform revealed where adoption would accelerate and where it would stall.
3. Energy capacity validation
The most critical and technically rigorous layer focused on validating whether energy systems could sustain projected demand.
Electricity forecasting
Using datasets from the U.S. Energy Information Administration (EIA 860 and EIA 861), models projected electricity generation capacity across sources:
- Natural gas
- Nuclear
- Renewables
- Hydro
These forecasts were benchmarked against the industry-standard Annual Energy Outlook 2025 to ensure alignment with real-world planning assumptions.
Natural gas forecasting
Parallel models were built for CNG production capacity using:
- Dry gas
- Supplemental gas
- Renewable natural gas
By combining these inputs, the team achieved forecasts within ~6.6% of EIA baseline projections — a level of accuracy that made the outputs credible for strategic planning.
From theoretical models to defensible insights
One of the biggest technical challenges was avoiding a common pitfall: unbounded extrapolation. Left unchecked, forecasting models tend to extend historical trends indefinitely, ignoring policy shifts, infrastructure limits, and market dynamics. The breakthrough came from anchoring all projections to EIA benchmarks.
By aligning outputs with Annual Energy Outlook 2025:
- Forecasts became defensible and industry-aligned
- Insights could be trusted by planning teams
- Outputs reflected how energy markets actually evolve, not just how data trends behave
This step transformed the platform from an experiment into a decision-grade system.
From theoretical models to defensible insights
One of the biggest technical challenges was avoiding a common pitfall: unbounded extrapolation. Left unchecked, forecasting models tend to extend historical trends indefinitely, ignoring policy shifts, infrastructure limits, and market dynamics. The breakthrough came from anchoring all projections to EIA benchmarks.
By aligning outputs with Annual Energy Outlook 2025:
- Forecasts became defensible and industry-aligned
- Insights could be trusted by planning teams
- Outputs reflected how energy markets actually evolve, not just how data trends behave
This step transformed the platform from an experiment into a decision-grade system.
The result: turning insight into opportunity
The platform delivered immediate strategic value by helping the customer navigate an unexpected market shift.
Policy changes accelerating EV adoption had created a scenario where:
- Historical trends were no longer reliable predictors
- Infrastructure constraints became critical decision factors
- New opportunities emerged in overlooked regions
With the new system, the customer could:
- Identify where energy supply exceeds demand and where to expand
- Detect regions where infrastructure gaps could limit adoption
- Align product strategy with realistic, data-backed adoption scenarios
One particularly powerful insight came from combining datasets in new ways:
“This has opened lots of eyes with surprise opportunities. By overlaying VIN numbers over natural gas pipelines, we’ve identified areas where they could be penetrating with more natural gas due to the supply, but they haven’t sold more.”
– Director of Advanced Connected Products and Digital Twins
This kind of insight moved the conversation from forecasting demand to shaping the market itself.
What’s next: expanding the intelligence layer
With a successful MVP in place, the platform is positioned to evolve further by integrating advanced datasets, including grid intelligence from Google initiatives like Tapestry. In the next phase, Onix will embed large language model (LLM) capabilities to automatically interpret complex dashboards and generate executive summaries for stakeholders. Furthermore, incorporating localized regulatory policies and incentives will refine forecasting accuracy, enabling the model to scale globally across new regions and diverse powertrain types. Ultimately, these capabilities will allow the customer to extend its predictive use cases into adjacent domains, such as data center energy planning and broader power systems strategy.
Conclusion: powering the future with data-driven confidence
The global power technology leader’s journey with Onix demonstrates that the future of mobility cannot be understood through vehicle trends alone; it requires a connected view of demand, infrastructure, and energy systems validated against how the real world operates. By building a predictive intelligence platform grounded in both data science and industry benchmarks, the customer has effectively moved from uncertainty to clarity.
This approach allows the organization to understand not just what will happen but precisely where and why, while simultaneously uncovering strategic opportunities hidden within infrastructure gaps. By creating a foundation for smarter, faster, and more confident strategic decisions, the customer is uniquely positioned to lead, rather than just respond to, the next era of energy and mobility in an industry defined by rapid transformation.
