Hey everyone,
I recently attended the Salesforce Energy & Utility Summit and had some great philosophical and technical discussions about where enterprise AI is actually heading. I wanted to share a few observations with the community, as the impression I walked away with points to a massive infrastructure boom, but also a very specific reality about the adoption curve.
Here are my key takeaways:
1. Heavy Industry & Closed-Loop Automation
The mining industry is already extensively using AI to run operations, specifically in the remote control of heavy equipment. A standout use case I saw was in utilities, utilizing drones for vegetation management. The drones aren't just taking photos; the AI is actively evaluating the photography and autonomously predicting and prescribing the physical actions needed on the ground. We are moving from AI as a "copilot" to AI as the primary operator in the field.
2. Enterprise Governance is Maturing Fast
One of the biggest hurdles for corporate AI has been security. I saw a great demonstration of MuleSoft orchestrating various AI agents, proving that tight governance is here. For example, they have hard architecture in place where a "general agent" cannot access or trigger what a specialized "finance agent" can do (like deleting an invoice). The software guardrails to make AI safe for enterprise deployment are solidifying.
3. The Adoption Reality: We Are Strictly at "Level 0"
Despite the advanced field operations, the actual corporate rollout in the utility sector is still in its infancy. Right now, companies are heavily focused on change management—training employees to use AI strictly as an internal "companion" (what they are currently calling Level 0). There is a hard boundary right now: teams are only confident using AI for internal, back-office processes. They are absolutely not ready or confident enough to let AI answer or face the customer directly. The immediate commercial focus is purely on internal efficiency.
4. The Macro Thesis: The AI Infrastructure & Resource Tax
This was the biggest lightbulb moment. To sustain this upcoming wave of industrial AI and agentic orchestration, we are staring down the barrel of a massive infrastructure boom. The physical constraints are staggering—during the discussions, it was highlighted that data centers are predicted to eventually consume up to 25% of water (for cooling) in heavily concentrated areas.
The Takeaway:
The next phase of the AI boom isn't just about software; it’s an energy, water, and hard-infrastructure play. As we look at the market, the picks-and-shovels play might literally be the companies providing the physical resources, cooling tech, and power grid upgrades required to keep these data centers running, while the software side slowly matures past "Level 0" internal use.
The content is mine structured by Femini