Everyone Is Wrong About AI’s Water Use - Here’s the Truth 💡

Welcome to another edition of Horizon AI,

AI's water consumption is often exaggerated or misunderstood, and it's placed in context with broader resource use and lifecycle analysis. Given how data centers and AI influence water and energy demands, we need to assess environmental claims critically rather than repeat misleading figures.

Let’s jump right in!

Read Time: 4.5’ min

Here's what's new today in the Horizon AI

• Chart of the week: The Global AI Arms Race Across Countries

• Google Translate Turns Any Headphones into Real-Time Language Translators

• AI Findings/Resources

• AI tools to check out

• Video of the week

TOGETHER WITH SYNTHFLOW

A Better Way to Deploy Voice AI at Scale

Most Voice AI deployments fail for the same reasons: unclear logic, limited testing tools, unpredictable latency, and no systematic way to improve after launch.

The BELL Framework solves this with a repeatable lifecycle — Build, Evaluate, Launch, Learn — built for enterprise-grade call environments.

See how leading teams are using BELL to deploy faster and operate with confidence.

Get the Guide

Chart of the week

The Global AI Arms Race Across Countries

• The chart highlights a widening AI divide that risks deepening economic inequality between high-income and developing countries. The rankings combine R&D, talent, infrastructure, policy, and public opinion, revealing a strong link between national income and AI strength.

Source

AI News

GOOGLE

Google Translate Turns Any Headphones into Real-Time Language Translators

Google is rolling out a major update to its Translate app that transforms everyday headphones into real-time translation devices, powered by its advanced Gemini AI. The beta feature, now available on Android, delivers live audio translations that retain natural voice tone and nuance, making multilingual conversations feel more seamless and intuitive.

Details:

• Users can now use the Google Translate app to hear real-time speech translations directly through any pair of headphones, no specialized hardware required.

• The system preserves the speaker’s tone, emphasis, and cadence, helping listeners follow conversations more clearly and retain emotional context.

• The feature supports over 70 languages, allowing live translation across diverse linguistic settings.

• Initially available on Android in the United States, Mexico, and India, with plans to expand to iOS and more countries in 2026.

• Gemini’s AI improves accuracy for idioms, slang, and nuanced expressions, delivering more natural, context-aware translations.

• To use, open the Translate app, pair your headphones, and tap “Live translate” to begin hearing translations as someone speaks.

Source

AI Findings/Resources

📱 How OpenAI used Codex to build Sora for Android in 28 days

⚡ 4 Perplexity Comet automations that save more time than any extension

  🏛 Global experts warn AI reshapes diplomacy and demands transparent, equitable tools and policies

🌋 Anthropics’ chief scientist says AI risks could lead to catastrophic outcomes.

📊 In-depth look at GPT-5.2 Pro capabilities, architectures, and performance trade-offs.

AI Tools to check out

🎼 Mozart AI: AI Music Creation Platform with Music Video Generation

🔗 Korgi: AI-built project boards powered by your productivity stack

🤖 Kaily: AI agent that sells, resolves, and automates daily tasks in a flash

🔎 Gemini Deep Research Agent: Autonomously plans, executes, and synthesizes multi-step research tasks.

🥗 PlanEat AI: AI turns your health goals into a 7-day menu & grocery list

Video of the week

The Truth About AI and Water Use and Why We Should Worry About Power Instead

The Core Conflict: There is a massive confusing gap in how people talk about AI’s water footprint. On one hand, OpenAI’s Sam Altman claims a single ChatGPT query uses just a tiny fraction of a teaspoon of water. On the other hand, reports project AI data centers could use over a trillion liters of water a year by 2028. Both numbers can be technically true depending on how you measure them, but both are misleading.

Why Measuring Is So Hard

• The "Hidden" Water: Sam Altman’s "teaspoon" figure likely only counts the water used at the exact moment you ask a question. It ignores the massive amount of water used to train the AI models (which takes weeks of running supercomputers) and to manufacture the computer chips.

• Different Types of Water: Not all water use is the same. Some data centers use fresh drinking water (municipal water) to cool servers. Others use recycled sewage water. Some use water that evaporates and disappears, while others borrow water from a river and put it back. Treating all these as the same thing confuses the issue.

• The Power Plant Connection: The biggest numbers often include the water used by power plants to generate the electricity that runs the AI. Power plants use huge amounts of water to cool steam, but most of that water is returned to the source, not "used up" like water sprinkled on a lawn.

The Bigger Picture

• Location Matters Most: Using water isn't inherently bad; using it where it is scarce is the problem. A data center in a wet region causes far less harm than one in a desert like Tucson.

• The Corn Comparison: To put things in perspective, the US grows corn to turn into ethanol fuel for cars. This process uses roughly 80 times more water annually than all the world's AI servers combined. While AI water use is significant, it is small compared to agriculture.

The Real Concerns The speaker argues that while water use is an easy thing to get angry about, it might not be the biggest problem.

1. Electricity is the bigger issue: The demand for power is growing much faster than water use, which could strain power grids and increase pollution.

2. The "AI Bubble": The most worrying risk might be economic. Tech companies are spending hundreds of billions building data centers based on the assumption that AI will change the world overnight. If that future doesn't happen, we may have wasted massive resources on infrastructure we didn't need.

Resource analysis is messy. It is easy to lie with numbers by either leaving out the training costs (to make AI look efficient) or including power plant water flow (to make AI look wasteful). The reality is somewhere in the complicated middle.

Meme of the week

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻