The move scales their joint operations in Sparks, Nevada, to seven times the original AI infrastructure density, providing a blueprint for how second-life batteries can power high-performance computing.

From Pilot to Scale: 7x Growth

The expansion follows a successful pilot program launched in June 2025. Initially, the project utilized four Crusoe Spark™ modular data centers. Following seven months of high performance, the companies are increasing the deployment to 24 modular data centers.

This growth is made possible by the hardware’s “modular” nature. Unlike traditional data centers that require years of stationary construction, modular units can be manufactured off-site and deployed in months.

Powering AI with Second-Life Batteries

A central component of this partnership is the use of “second-life” electric vehicle (EV) batteries. When EV batteries are no longer optimal for automotive use, they often retain significant capacity for stationary energy storage.

Redwood Materials integrates these repurposed batteries into a 12-megawatt (MW) / 63-megawatt-hour (MWh) microgrid. This system, combined with on-site solar power, provides the energy required to run Crusoe’s AI-optimized GPUs. The orchestration of these batteries is handled by Redwood’s “Pack Manager” technology, which ensures steady power delivery for the intense workloads required by AI model training and inference.

Reliability and Performance Metrics

A primary concern with renewable-powered microgrids is “uptime”, the percentage of time the system is operational. The press release highlights several key performance indicators from the initial seven-month period:

99.2% Operational Availability: The microgrid exceeded reliability expectations while running on renewable sources and battery storage.

99.9% Total Uptime: By leveraging the traditional power grid as a backup source, Crusoe Cloud maintained a nearly constant state of operation.

Supply Chain and Sustainability

The partnership addresses two of the most significant bottlenecks in the current AI boom: energy consumption and deployment speed.

Sustainability: By using recycled materials and on-site renewable energy, the “AI factory” model reduces the carbon footprint associated with massive data processing.

Predictability: The ability to scale in months rather than years allows AI providers to meet the rapidly fluctuating demand for compute power.

As the demand for intelligence grows, the convergence of innovative energy storage and modular infrastructure—as demonstrated by Crusoe and Redwood Materials—offers a potential path forward for sustainable and rapid industrial scaling.

The post Crusoe and Redwood Materials Expand Strategic Partnership appeared first on Logistics Viewpoints.

A recent briefing with Lytica highlights a shift in procurement from opaque negotiation toward statistically grounded pricing intelligence.

Procurement has long operated with an imbalance of information.

Suppliers understand pricing across customers, volumes, and market conditions. Buyers rely on internal history, limited benchmarks, and negotiation experience to determine whether a price is competitive. In categories such as electronic components, this gap is amplified by volatility and limited transparency.

The result is consistent. Different companies, and often different divisions within the same company, pay materially different prices for the same component.

Lytica is attempting to address that condition.

From Transaction Data to Market Intelligence

Lytica’s platform is built on anonymized buyer transaction data aggregated across a network of companies. This creates a continuously updated view of pricing across suppliers, regions, and time.

This is not modeled data or survey input. It reflects observed market behavior.

That distinction allows procurement teams to assess pricing against a broader market reference:

Where are we overpaying

How do suppliers price across customers

What does competitive pricing look like

This represents a move from internal spend analysis to external market intelligence.

From Benchmarking to a Pricing Discipline

The more important development is how this data is modeled.

Lytica treats pricing as a measure of competitiveness rather than a fixed value. Prices exist within a distribution shaped by real transactions. Each company occupies a position within that distribution.

This enables a more structured evaluation of procurement performance:

Prices can be ranked relative to the market

Outliers can be identified and examined

Expected price ranges can be estimated using observed data

The question shifts from “Is this price good” to “How competitive is this price relative to the market”

This introduces a more disciplined approach to procurement performance.

Quantifying Leverage in Negotiation

Once pricing is modeled this way, negotiation becomes more structured.

Procurement teams can enter discussions with:

Target pricing ranges based on transaction data

Evidence of variance across comparable buyers

Supplier-specific pricing patterns over time

This replaces qualitative positioning with data-backed arguments.

The result is more consistent outcomes and shorter negotiation cycles.

From Data to Decision Support

The next step is applying this dataset in operational workflows.

As outlined in modern supply chain architectures , AI systems become more useful when grounded in domain-specific data and applied with context.

In this case, systems can:

Identify deviations from competitive pricing levels

Estimate expected pricing ranges based on observed transactions

Generate supplier-specific negotiation guidance

Monitor pricing performance over time

These outputs are typically delivered as structured guidance for sourcing teams.

The Role of Context and Retrieval

The effectiveness of this approach depends on how data is accessed and retained.

Retrieval-based architectures allow systems to reference current transaction data when generating recommendations. Context-aware systems retain supplier history, pricing behavior, and prior outcomes across decision cycles.

This supports continuity in decision making rather than isolated analysis.

Positioning in the Stack

Lytica does not replace ERP or sourcing platforms. It operates as an intelligence layer above them.

This reflects a broader shift:

Systems of record manage transactions

Systems of execution manage workflows

Systems of intelligence guide decisions

Over time, as confidence in recommendations increases, this layer is likely to become more integrated into execution.

The Bottom Line

Lytica reflects a shift in procurement.

Pricing is moving from opaque negotiation toward structured, data-based market positioning.

This changes how procurement operates:

From internal benchmarks to external reference points

From periodic sourcing to continuous evaluation

From intuition to structured decision support

In more volatile supply environments, this type of capability becomes increasingly relevant.

Organizations that adopt it early will have a clearer understanding of their market position and a more consistent approach to improving it.

The post Lytica and the Emergence of a Pricing Science Layer in Procurement appeared first on Logistics Viewpoints.

AI is no longer confined to planning systems and dashboards. It is moving into the execution layer of the supply chain, where decisions are made in motion, not after the fact.

For the past decade, most AI investment in supply chains has focused on forecasting, planning, and analytics. These systems improved visibility and supported better decisions, but they remained upstream. Warehouses, fleets, ports, and production lines continued to operate with limited real time intelligence.

That separation is now collapsing.

A new phase is emerging where AI is embedded directly into physical operations. Systems are no longer just recommending actions. They are beginning to sense conditions, coordinate responses, and execute decisions across the network.

This shift has material implications for cost, service levels, and resilience. It also changes where value is created and who controls it.

The Shift from Insight to Execution

Most supply chain AI to date has been advisory. It has answered questions such as:

What will demand look like next month

Where should inventory be positioned

Which supplier carries the lowest risk

These are important questions, but they sit upstream from execution.

The next wave moves downstream. It focuses on questions such as:

What should happen to this shipment right now

How should this route change given current conditions

Which order should be prioritized inside the warehouse

These decisions are continuous and time sensitive. They cannot wait for batch planning cycles or manual intervention. As AI moves into execution, the cadence of decision making shifts from periodic to continuous. That is where the real operational leverage sits.

The Supply Chain Is Becoming a Network of Active Nodes

Physical supply chains are being instrumented. Vehicles, containers, facilities, and even individual assets are becoming data generating nodes.

Each node produces signals about location, status, constraints, and performance. More importantly, these nodes are no longer passive.

They are beginning to participate in decision making.

A truck is no longer just executing a route. It is part of a system that can:

Adjust routing based on congestion and delivery windows

Coordinate arrival times with warehouse capacity

Trigger downstream inventory decisions

A warehouse is no longer just processing orders. It is dynamically adjusting labor allocation, slotting, and picking sequences based on incoming conditions.

This changes the structure of the supply chain from a linear process to a responsive network.

Coordination Becomes the Core Problem

As intelligence moves into physical operations, the primary challenge is no longer prediction. It is coordination.

Optimizing one function in isolation delivers limited value. A perfectly optimized route has little impact if the receiving facility cannot process the shipment. Inventory decisions fail if transportation and supplier realities are not aligned.

What matters is how decisions interact across the system.

This is where many current deployments fall short. They optimize within silos. The next phase connects those silos.

Execution systems are beginning to coordinate across:

Transportation and warehousing

Procurement and inventory

Order management and fulfillment

The result is not just faster decisions. It is better system level outcomes.

The Compression of Decision Cycles

One of the clearest signals of this shift is the compression of decision cycles. Traditional supply chains operate on defined rhythms. Daily planning runs. Weekly forecasts. Monthly reviews. Physical execution does not operate on those timelines. Disruptions occur in minutes. Conditions change continuously. Opportunities are fleeting.

As AI moves into execution, decision cycles compress from hours and days to seconds and minutes.

This has three direct effects:

Reduced latency between signal and action

Fewer manual interventions

Increased ability to absorb disruption without escalation

The organizations that adapt to this cadence will operate with a structural advantage.

Where Value Is Moving

As AI enters the physical layer, value is shifting. Historically, value concentrated in planning systems and enterprise platforms. These systems aggregated data and produced recommendations. Now, value is moving toward the execution layer, where decisions are acted on.

Three areas stand out:

1. Real time orchestration
The ability to coordinate decisions across transportation, warehousing, and inventory in real time.

2. Embedded intelligence in assets
Vehicles, automation systems, and edge devices that participate in decision making.

3. Network level visibility tied to action
Not just seeing what is happening, but acting on it immediately.

This has implications for technology providers, operators, and investors. Control points are shifting.

What Leaders Should Watch

This transition is underway, but uneven. Most organizations are still early.

There are several signals worth tracking.

Execution level use cases moving to production
Look for systems that are not just advising planners but actively influencing routing, picking, allocation, and scheduling.

Tighter integration across systems
Disconnected tools will not support this model. Integration across TMS, WMS, and upstream systems becomes critical.

Rise of real time data pipelines
Batch processes will not support continuous decision making. Event driven architectures will.

Shift in organizational roles
Planners move from direct decision making to oversight and exception management.

Vendor positioning around orchestration
The most important platforms will not be those that optimize a single function. They will be those that coordinate across the network.

The Risk of Standing Still

The risk is not that AI fails to deliver. The risk is that competitors operationalize it first. A supply chain that can sense and respond in real time will outperform one that relies on delayed information and manual coordination.

The gap will not be incremental. It will be structural. Faster response times, better asset utilization, fewer disruptions, and higher service levels compound quickly. Organizations that remain in a planning centric model will find themselves reacting to a system that is already moving.

The Bottom Line

AI in the supply chain is no longer about better forecasts or improved dashboards. It is about execution.

As intelligence moves into the physical layer, supply chains become more responsive, more coordinated, and more resilient. Decisions happen continuously, across the network, not in isolated systems.

The leaders who recognize this shift early and align their architecture, data, and operating model accordingly will define the next generation of supply chain performance.

The post AI Is Moving Into the Physical Supply Chain: What Leaders Should Watch appeared first on Logistics Viewpoints.