Coca-Cola is turning to AI to save the global orange supply

Your morning glass of orange juice is in serious danger, and Coca-Cola is turning to artificial intelligence for a rescue mission.

The beverage giant behind Minute Maid has teamed up with MIT, OpenAI, India’s Tata Group, and South Korea’s SK Telecom to launch the MIT Generative AI Impact Consortium. Their mission is to apply advanced AI to urgent, real-world problems.

Their very first initiative has a blunt name: Project "Save the Orange."

A citrus apocalypse in slow motion

The enemy here is citrus greening, an incurable bacterial disease that has devastated groves worldwide for two decades.

First detected in Florida over twenty years ago, the disease has infected most of the state's orange trees. Combined with severe hurricanes, Florida's citrus output has collapsed by more than 90% since 2005.

The numbers from the USDA paint a grim picture. The United States once produced 45% of the world’s oranges, but today accounts for a mere 5%. For 2025, the USDA projects U.S. orange production at just 2.5 million tons, down from 5.5 million tons a decade ago.

The crisis has now engulfed South America. In 2024, citrus greening hit nearly 45% of the vital Citrus Belt in Brazil, the world's top exporter.

Coca-Cola issued a blunt warning: without radical breakthroughs in disease detection, grove management, and treatment, the planet's orange supply could virtually disappear within the next 25 years. To build Project Save the Orange, the company is partnering with Brazilian growers association Fundecitrus and agritech firm Invaio Sciences.

Moving past AI marketing stunts

Coca-Cola is no stranger to artificial intelligence, though past efforts leaned heavily on novelty.

Back in 2023, the brand rolled out Coca-Cola Y3000, a limited-edition soda billed as the first flavor co-created with human and artificial intelligence. It has also used generative models for marketing campaigns, dynamic ads, and basic logistics forecasting.

Project Save the Orange is a completely different beast. Instead of generating futuristic can designs, AI models will now tackle complex life sciences, aiming to spot infections earlier and assist biological interventions in the field.

Why it matters

We think this coalition represents a refreshing shift in how big enterprise views generative AI.

For the past two years, consumer brands mostly treated generative models as a cheap gimmick for viral ad campaigns. Using foundation models to combat a devastating plant pathogen shows AI moving directly into mission-critical physical supply chains. If you cannot secure oranges, Minute Maid simply ceases to exist.

For everyday consumers, this battle directly dictates whether breakfast staples remain affordable or turn into luxury items driven by chronic crop failure.

The big question mark is execution. Machine learning can dramatically accelerate molecular research and pattern spotting in satellite imagery, but biological ecosystems do not operate on digital timelines. Field trials take years, and curing an infected tree is fundamentally harder than rendering an image. Coke's 25-year countdown is running, and technology will need to deliver tangible agricultural results fast.