Rather than turning to power-hungry AI tools to tackle every business challenge, Intel urges leaders to be selective and focus on their highest-value corporate priorities. “You have to understand the business goals, the business incentives and what you're trying to drive,” says Motti Finkelstein, corporate vice president and chief information officer at Intel.
Once the areas where AI can make the biggest impact are identified, such as refining manufacturing processes or improving data center performance, Intel leverages optimized models, software and hardware to help develop and run more energy-efficient AI processes:
Model Behavior
AI models are trained on data to recognize patterns or make decisions independently. They act as the brain of AI-enabled software. The more complex the model, the more computing power and energy it requires. Intel tools and support are available to reduce model size through compression techniques such as quantization.
Customizable Code
Intel says that up to 18% of energy consumption in a data center is due to software inefficiency—as software is often built for function, not efficiency. Intel offers software optimizations for many applications and frameworks that can reduce power and increase performance.
Enhanced Hardware
To avoid unnecessary compute and energy usage, you want the hardware to match the needs of a given task,” says Finkelstein. For example, Intel Xeon Processors with built-in accelerators are used in data center servers for demanding AI workloads. Huffstetler says that the processors have been refined so much that just three fifth-generation systems can replace 50 first-generation systems.
“
From cultivating crops to coordinating calls, businesses across industries use Intel’s solutions to hone a previously unimaginable competitive edge by reducing waste.
UNLEASHING INNOVATION
FARMING POWERED BY 2,000 SENSORS
Every year, $1 trillion worth of produce goes directly from farms to landfills due to factors including weather, pests, market volatility and supply chain challenges. Nature Fresh Farms, a Canadian greenhouse agriculture company, uses Intel technology to run AI-powered processes that reduce food waste and produce up to 10 times the crop yield per acre they’d expect to see with traditional farming.
Thousands of sensors capture data across the company’s vast greenhouses. Software optimized using the Intel OpenVINO Toolkit and running on Intel Xeon Processor-powered servers rapidly processes and evaluates this data. The insights assist farmers in refining many aspects of the growing and harvesting process.
Click To Explore How Nature Fresh Farms Automates Agriculture
A leading self-driving technology company uses machine learning to analyze data from vehicle sensors and cameras to power advanced driver-assistance systems (ADAS) and autonomous driving technologies.Vast amounts of data are processed in the cloud to rapidly create centimeter-accurate road maps, which feed into a global bank of driving paths. The company chose the AI-driven solution Intel® Tiber™ App-Level Optimization to improve the efficiency of its applications and reduce power consumption.
OPTIMIZING OPERATIONS
MAPPING THE WORLD FOR SAFER DRIVING
How A Navigation System Entered The Fast Lane
Greater Agility
Because App-Level Optimization is autonomous and does not require human intervention, DevOps teams can focus their time and resources on core product research and development.
Accelerated Achievement
Patented algorithmic models continuously identify and remedy data bottlenecks, leading to 45% faster mapping completion times and an annual carbon reduction of approximately 30,000 kg.
Rapid Rollout
Minutes after deployment, App-Level Optimization began analyzing and learning the company’s dataflow and resource usage patterns and was ready to be activated within a week.
Proactive Failure Alerts
Intel uses sensors to monitor factory equipment, and AI analyzes the data to predict problems and detect potential issues like leaks. If a machine malfunctions, it can be immediately halted. This system saves millions of dollars by reducing waste and machine downtime.
Powerful Digital Twins
Intel uses detailed virtual models of production line equipment that mimic the behavior of their real-world counterparts. These twins allow the company to identify efficiencies and simulate any changes before implementing them.
Automated Quality Control
High-resolution cameras take multiple images per second across the semiconductor production line, which a machine-learning model analyzes. If defects are spotted, an alarm sounds. This system saves Intel $2 million annually, reduces waste and frees engineers from conducting manual inspections.
How Intel Embraces Its Own Sustainability VALUES
PERFECTING PERFORMANCE
A VISION FOR PRODUCTION LINE PRECISION
Intel is deploying AI to make itself more efficient and eco-friendly. “We take a holistic view across product development, manufacturing and sales with the goal of improving productivity, quality and resource utilization by 30-40% over the next two years,” Finkelstein says.
“We want the whole world to build sustainable AI infrastructure, but also to use AI for sustainability.”
Chief Product Sustainability Officer, Intel
Creating The
Ideal Climate
Critical variables such as temperature and humidity are recorded and rapidly turned into insights that can, for example, trigger the closing of energy curtains (greenhouse window coverings that block the sun).
Seeking And Destroying Pests
Irrigating With Precision
Picking PerfectProduce
Tracking Every
Plant
Devices such as cameras can help identify when crops are ripe and ready to be picked. Nature Fresh Farms then uses AI-powered sorting to quickly grade and pack produce that arrives on shelves within 24-48 hours.
Crops are watered using a sophisticated
AI system to determine the exact timing and amount needed. The process is 90% more water-efficient than traditional farming methods.
The farm has experimented with AI models that analyze data to detect signs of pest activity on plants, like wilting leaves or low water uptake. To avoid pesticides, ladybugs are used to respond to nuisances like aphids.
A single plant generates 12 megabytes of data during its lifecycle, which allows the company to forecast how much produce each facility will harvest over the coming days and weeks.
Click here to learn more about unlocking AI-powered sustainability in your organizations.