By 2025, the battleground for AIGC has shifted from text to visuals. If ChatGPT has enabled everyone to “write,” then Sora 2 and Nano Banana Pro are empowering everyone to “make movies.” More importantly, this visual revolution demands hundreds or even thousands of times more computing power, storage, and energy consumption than text generation, opening up an entirely new super track.

I. Overview of Developments in the Visual AI Industry

• OpenAI Sora 2: The 'GPT-3.5 Moment' for Video Generation

At the end of September 2025, OpenAI quietly launched Sora 2, which supports up to 60 seconds of 1080p video with synchronized audio and dynamic generation compliant with physical laws. Data speaks volumes: within five days of its iOS app release, Sora topped the U.S. App Store free charts with one million downloads, achieving a first-week download volume of 627,000.

This explosive growth was achieved despite Sora's extremely high barriers to entry, as it is accessible only via iOS devices and an invitation code, leaving the vast majority of potential users still locked out.

• Google Nano Banana Pro: Dimensionality Reduction Attack with One-Click 4K Image Generation

On November 20th, $Alphabet-C (GOOG.US)$ Google released Nano Banana Pro, an image generation and editing model built on the Gemini 3 Pro system. In addition to improved image quality, Nano Banana Pro incorporates a new 'reasoning core,' enabling the model to analyze user requirements, design compositions, and check logic like a professional designer before rendering begins.

Google integrated the reasoning capabilities of Gemini 3 Pro and its world knowledge base into Nano Banana Pro. The key difference between the Pro version and the standard Nano Banana lies in its professional-grade upgrades, offering native 2K rendering and 4K ultra-high-resolution enlargement technology. This means that images output by the Pro version can be directly used for professional printing and commercial purposes.

Furthermore, the cost of generating 4K images with Nano Banana Pro has been reduced tenfold, significantly enhancing its feasibility for commercial applications. The day after its release, $Alphabet-C (GOOG.US)$ The stock price surged over 3%, elevating its market value to the third largest in the U.S. stock market.

Not long ago, Google CEO Sundar Pichai revealed that, driven by Nano Banana, the number of monthly active users of the Gemini app skyrocketed from 450 million to 650 million in just one quarter.

• With the upcoming earnings report, can Adobe stage a comeback?

In the past couple of years, the market has been pessimistic about $Adobe (ADBE.US)$ “It’s going to be outcompeted by Midjourney.” Investors believed that emerging AI creative tools, with lower costs and more user-friendly interfaces, were attracting a large number of individual and small-to-medium enterprise users, putting Adobe under dual pressures of subscription models and high pricing strategies. Adobe's stock price has also been on a continuous decline since the second half of 2024.

To keep up with the wave of the AI era, $Adobe (ADBE.US)$ Adobe launched Firefly. According to the 2024 financial report, the generative AI tool “Firefly” introduced by Adobe, along with its deep integration into Creative Cloud and Acrobat, has indeed attracted new users, enhanced customer stickiness, and helped achieve sustained growth in recurring customer revenue.

The company has deeply embedded Firefly into flagship applications like Photoshop while launching standalone apps, forming an all-in-one AI suite model. Adobe is set to release its earnings report after the U.S. stock market closes on December 10th, at which point the financial data will reveal whether Adobe is a laggard in the AI era.

II. Visual AI Amplifies Demand Hundreds of Times Over—Which Links in the Chain Are Profitable?

From Sora 2 to Nano Banana Pro, we are standing at the cusp of a visual AI explosion. This is not a breakthrough in a single technology but rather a complete industrial chain opportunity spanning from foundational chips and cloud services at the bottom layer, to intermediate-layer models and tool software, and finally to the upper-level content ecosystem.

1. AI Computing Chips: $NVIDIA (NVDA.US)$ 、 $Broadcom (AVGO.US)$ 、 $Advanced Micro Devices (AMD.US)$

Whether training a behemoth like Sora 2 or running inference on Nano Banana Pro, high-performance AI chips are indispensable. $NVIDIA (NVDA.US)$ With its leading CUDA ecosystem and hardware dominance, NVIDIA remains the undisputed cornerstone.

However, the tsunami of market demand has also created vast alternative and incremental opportunities for $Advanced Micro Devices (AMD.US)$ MI series accelerators, $Broadcom (AVGO.US)$ and customized AI chip solutions from other providers. Advanced Micro Devices (AMD) leads in video inference cost-performance with its MI350 series, while Broadcom secured a major contract for Google's TPU with its custom ASICs and optical interconnect chips.

2. Cloud Computing Platforms: $Microsoft (MSFT.US)$ 、 $Alphabet-A (GOOGL.US)$ 、 $Amazon (AMZN.US)$ 、 $CoreWeave (CRWV.US)$

The need for model training and large-scale API calls has made cloud services a necessity. Microsoft Azure, Google Cloud, and Amazon AWS, with their global infrastructure, form the top tier. AWS offers over 200 fully-featured services ranging from computing, storage, and databases to IoT and quantum computing, capturing the largest market share. As of Q3 2025, AWS holds approximately 31% of the market share, while Azure and Google Cloud hold about 25% and 11%, respectively.

And$CoreWeave (CRWV.US)$ Starting from the early days of cryptocurrency mining, CoreWeave accumulated GPU resources and strategically pivoted to become a cloud service provider. Deeply integrated with NVIDIA, it has grown into a unicorn by leveraging flexibility and performance advantages amidst competition from industry giants. Unlike traditional public cloud providers offering generalized services, CoreWeave is specifically designed and optimized for large-scale, high-intensity AI workloads.

In Q3 2025, CoreWeave reported revenue of $1.365 billion, a significant year-over-year increase of 134%. Additionally, CoreWeave disclosed a backlog of orders totaling $55.6 billion, underscoring the strong long-term demand for AI cloud computing capabilities.

3. Storage: $Micron Technology (MU.US)$ 、 $SanDisk Corp (SNDK.US)$ 、 $Western Digital (WDC.US)$

Training models like Sora 2 and Nano Banana Pro requires unprecedentedly large datasets of videos and images. These datasets often reach petabyte (PB) or even exabyte (EB) levels and demand high-speed storage for frequent access. This directly drives the demand for enterprise-grade solid-state drives (SSDs) and high-speed storage solutions.

When generating videos, especially 4K/8K content, models need to perform complex intermediate calculations and frame caching. For instance, generating a six-second video may involve the temporary storage of hundreds of intermediate rendered frames, imposing extremely high requirements on storage bandwidth and latency. This pushes the demand for high-density DRAM and high-speed SSDs.

$Micron Technology (MU.US)$ 、 $Western Digital (WDC.US)$ 、 $SanDisk Corp (SNDK.US)$ Leading storage companies will directly benefit from the long-term demand boost brought by the era of visual AI.

4. Visual AI Generation:

$Alphabet-A (GOOGL.US)$ Equipped with the Nano Banana Pro image model and multiple video generation technologies, its core competitiveness lies in the deep integration of multimodal large models, global distribution channels, and robust engineering capabilities. $Figma Inc (FIG.US)$ Figma AI, an AI design tool offered by Figma, can automatically generate UI components and layouts. Its competitive edge lies in its extensive designer community ecosystem and closed-loop real-time collaboration features. $Adobe (ADBE.US)$ The 'Firefly' series of models are deeply integrated into the entire suite of products, including Photoshop and Premiere. The key competitive advantage is Adobe's monopolistic position in creative software and seamless embedding within professional workflows.

A 3D interactive content development platform $Unity Software (U.US)$ and $Meta Platforms (META.US)$ At Unite 2025, a full suite of toolchains for VR development was jointly showcased, with a focus on lowering the development threshold and improving efficiency.

5. Visual social media: $Meta Platforms (META.US)$ 、 $Alphabet-A (GOOGL.US)$ 、 $Reddit (RDDT.US)$

Social media platforms are set to undergo another content supply revolution, with user-generated content (UGC) evolving from images and text, as well as short dramas, into high-quality AI-generated dynamic videos.

Tools such as Sora 2 will significantly lower the threshold for video creation, enabling a vast number of users to generate high-quality, highly engaging short videos and images. This directly benefits $Meta Platforms (META.US)$ (Facebook, Instagram), $Alphabet-A (GOOGL.US)$ (YouTube, Shorts), and $Reddit (RDDT.US)$ , providing richer and more cost-effective platform content that enhances user engagement and increases session duration.

We may be on the eve of an 'AI video explosion.' From Sora 2 to Nano Banana Pro, humanity now possesses, for the first time, the ability to 'directly make films with thoughts,' and this revolution is only just beginning.

Looking to pick stocks or analyze them? Want to know the opportunities and risks in your portfolio?For all your investment-related questions, just ask Futubull AI!

Editor/Doris