[ad_1]
Viral as Nano Banana, the tool has garnered record impressions. Josh Woodward, vice-president and global head of Gemini, tweeted that by 9 September, Google’s AI platform had seen over 23 million new user registrations and generated over 500 million images in two weeks.
Experts say Google’s viral image-generation platform could mark the beginning of mass-scale enterprise AI adoption—more so than OpenAI’s Studio Ghibli-style experiment earlier this year.
Mint explains the length and breadth of its potential impact.
How does Nano Banana work?
Put simply, Nano Banana is an AI model solely dedicated to text-to-image generation. While its broader model, Gemini 2.5, works across multiple input methods such as text, image, video and voice, Nano Banana was trained solely on image-based datasets—including image types, textures, 3D modelling, and rendering of spatial information to understand aspects such as the direction of shadows, the depth of a room, proportions of a human body, and more. It works in the same way as any other AI model, and can be accessed by simply choosing the Gemini 2.5 Flash Image—which is the technical name for it. Currently, the service is available to all users for free.
What explains the hype?
Until now, image-generation tools—such as ChatGPT powered by OpenAI’s multipurpose GPT-4o—have largely focused on showing how text prompts can instantly create images. Similarly, smartphone makers have also integrated image-generation applications to show users how the technology would work—this includes Apple’s Image Playground in iPhones, and Google’s Pixel Studio in its own smartphone line-up.
The Nano Banana image model, via the Gemini app, is the first one being used more for image editing and juxtaposition than for generating images.
Technical data from Google suggests that the model’s ‘context window’, or the amount of information you can feed it for it to understand your prompt, is over 1 million tokens. In simpler terms, you can add up to 3,000 images or files in one prompt—to explain the exact kind of image that you want to see generated. You can also ask the platform to combine two images in a specific form—for instance, uploading a normal photograph of a room and a piece of furniture, and asking Gemini to show the results of how such furniture placement would revamp the room.
Would this affect businesses and jobs alike?
Experts believe so. Four analysts, consultants, and advisors that Mint spoke with said businesses are already experimenting with Gemini’s latest image generator and how such a tool may reduce business expenses. Companies are also looking to understand how to frame the right prompt, which experts said may lead to prompt engineers returning to the industry.
In a report earlier this year, consulting firm McKinsey & Co. said generative AI can add up to $4.4 trillion in economic value each year—a figure that may double if generative AI models are built into software platforms that enterprises use today. Further, nearly 40% of this impact is expected to play out in customer operations, marketing, and sales support.
Gartner data from 2024 projected customer service and support as an industry to be worth nearly $50 billion. Similarly, the broad umbrella of marketing and sales is already worth multiple trillions of dollars.
Experts said each of these industries uses creative design and image services extensively. It is home to over four million employees across outsourced operations or call centres. More executives work across sales and marketing roles within companies—a figure that is more difficult to estimate.
Further, engineering, interior design and other industries are more squarely dedicated to design. Each of these industries uses tools and employs people to create designs for customers. All of this stands to be disrupted, with Gemini offering a plug-and-play platform for all kinds of design operations. For instance, Nano Banana’s understanding of 3D depth can help interior designers swap expensive 3D rendering tools for AI platforms, which are faster and, for now, less expensive as well.
What about privacy and information security concerns?
Each of these concerns is legitimate. Google discloses in the Gemini Apps Privacy Hub that information saved in a user’s account is shared for training its models. These include “what you say to Gemini applications, what you share—like files, videos, photos and page content, transcripts, and recordings of voice interactions with Gemini Live, usage feedback, as well as names and custom instructions”.
Other information collected includes content generated by the Gemini app platform, information from connected Google applications such as the Chrome web browser, YouTube search history, devices you own, as well as call and message logs, contacts, installed apps on your phone, content on your screen, and your location information.
Google discloses that this vast trove of data can and will be used to “maintain and improve services, develop new services, personalize services, measure performance, and protect Google, our users, and the public”.
While privacy advocates and users alike have raised concerns, experts advise that users can take some steps to restrict privacy and security concerns, such as turning off Google’s ability to use your audio and voice information recorded by Gemini, as well as turning off or deleting ‘Gemini app activity’ so that Google does not store copies of your information forever.
They also add that the company will likely offer enterprise-grade custom security protocols in the long run, once business use cases start to develop. These would include the ability for businesses to run Nano Banana offline or within the confines of a company’s own servers, thus restricting who would have access to sensitive data across the internet.
If the services are free, how would Google earn?
The services will remain free until mass-scale adoption. Further, businesses that use design tools at scale will inevitably need to pay to access the full suite of services. Google, on this note, already has its ‘AI Pro’ and ‘AI Ultra’ tiers in its ‘One’ subscription plan, which cost ₹1,950 and ₹24,500 per month, respectively.
Customers, too, will need to pay to use the image-generation service beyond the free tier, which restricts users in terms of the maximum volume of content they can generate, the frequency at which the tools can be used, and downtime caused by peak usage hours.
How is the company’s stock reacting?
The stock of Alphabet, Google’s parent firm, on the US stock exchange Nasdaq, has risen 21% since the rollout of Nano Banana. Others, too, are seeing lofty valuations.
Earlier this week, China’s Bytedance launched the latest version of its image-generative AI platform, SeeDream 4.0. US-based investor Fidelity, which holds a stake in TikTok’s parent firm, has valued the company at $385 billion, making it larger than Korean electronics conglomerate Samsung, Dutch semiconductor giant ASML, French luxury firm LVMH, and US industrial group General Electric, among others.
[ad_2]
Source link
