As a hot AI company, Open AI has once again secured a huge round of funding that is enough to make its competitors green with envy. On October 2, Open AI announced the successful completion of a $6.6 billion funding round, which not only compensated for the losses incurred from the money-burning R&D process but also provided financial support for the next round of competition.

Despite becoming a trillion-dollar unicorn, Open AI's anxiety persists. The high costs of R&D, personnel, and the transition to a profitable company have all become whip urging Open AI to accelerate its pace.

Fortunately, while firmly committing to becoming a profitable company, Open AI has already begun to lay out a grand plan for its next generation of super AI products.

Open AI doubles down on products with a two-pronged approach

Within a week of announcing its funding plan, Open AI has been actively making moves:

Firstly, at the second annual Developer Conference on October 1, Open AI unveiled multiple practical API calling plans. A day later, they officially announced the completion of the $6.6 billion funding round, catapulting them into the trillion-dollar unicorn club. Without any prior announcement, they also released a new product called Canvas, which is now open to Plus members.

With another successful funding round, Open AI is eager to pursue a two-pronged approach, capturing both the B-end and C-end markets simultaneously.

On October 1, at the second annual Developer Conference, instead of launching a product, Open AI provided developers with tool-based benefits by releasing four APIs: Real-time Speech API, Visual Fine-tuning API, Prompt Caching API, and Model Distillation API.

For subscribers, Open AI has also provided meaningful feature updates.

Early in the morning on October 4, Open AI officially released Canvas, an add-on product feature built on the GPT-4o model that enables collaboration with AI to complete writing and coding interfaces. It is currently available for testing by Plus and Team subscribers of ChatGPT.

In terms of interface design, Canvas abandons the traditional model and uses a blank interface to accommodate various AI work requirements. "My vision for the ultimate AGI interface is a blank canvas," said Karina Nguyen, a member of the Canvas development team.

Unlike previous multi-turn dialogue interaction forms, Canvas allows users to highlight parts of the creation interface that need adjustments, which are then fine-tuned locally by AI. If dissatisfied, users can restore previous versions using the "Undo" button.

To cater to different needs, Open AI has also designed specific shortcuts to help users increase efficiency. Taking writing as an example, Canvas has five built-in shortcuts: Suggest Edits (provides revision suggestions), Adjust Length (expands/shortens content), Change Reading Level (from kindergarten to graduate level), Final Polish (checks grammar, expression, and consistency), and Add Emoji (inserts emoji expressions into text).

Optimized editing function for highlighted sections Source: Open AI

Sam Altman himself seems particularly satisfied with the ability to add emoji expressions, even going so far as to post a poll on X asking, "Is adding emojis the best feature Open AI has ever released?"

Sam Altman only provided "yes" options for users to choose from, demonstrating his satisfaction with the feature Source: X

Canvas is also committed to becoming a more user-friendly and intuitive product. In addition to providing new interaction methods and convenient command functions, Open AI has also introduced intelligent recognition: when ChatGPT detects a potentially helpful scenario, Canvas automatically opens. Alternatively, typing the command to launch Canvas in the chat window will also redirect to the corresponding interface.

It is worth noting that compared to previous Open AI products, Canvas has truly addressed user pain points in terms of product feature design and interaction experience.

In terms of feature design, Canvas offers shortcuts that significantly enhance work efficiency, with some features receiving positive feedback from users. For example, the "Code Review" feature for programming uses AI capabilities to automatically check and modify issues in code.

In terms of interaction experience, Canvas stores text and code, which do not require contextual memory, in a separate blank interface and enhances modification flexibility through partial selection. Coupled with the smooth switching of the shortcut icon bar on the right side and the transition animation effects of content generation, Open AI's product details are becoming increasingly rich.

Behind the specific productization efforts lies the challenge that ChatGPT faces in limited paid user growth.

According to The New York Times, as of August, ChatGPT had 350 million monthly active users. However, despite this vast user base, the number of subscribers was only 10 million. Relying solely on current paid users to support substantial revenue targets is challenging.

Similar to the challenges faced by domestic chatbots like Doubao, Wenxiaoyan, and Kimi, ChatGPT's user inquiries in general scenarios are limited, capturing only a portion of the general search market.

To further enhance willingness to pay and average revenue per user, suitable products must be identified for specialized scenarios. This is why OpenAI is focusing on both the B-end and C-end markets.

Accelerating commercialization: Productization is key

With back-to-back updates and a successful funding round, Open AI is currently in the spotlight.

On October 2, BBC reported that Open AI had successfully completed its latest funding round of $6.6 billion, led by Thrive Capital (Goldman Sachs Capital) with participation from tech companies like SoftBank, NVIDIA, and Microsoft. Currently, Open AI is valued at $157 billion, solidifying its position as a trillion-dollar unicorn.

However, beneath the glory lies Open AI's ambition to become profitable and its disclosed revenue figures. Constrained by previous investment agreements, Open AI is considering transforming into a profitable company.

Bound by the terms of its agreements, Open AI's revenue generation plans must be accelerated. According to The Washington Post, insiders revealed that the terms of the new funding include a provision that allows investors to convert their shares in the company into 9% interest-bearing debt if OpenAI fails to complete its transformation within two years.

Media reports indicate that Open AI's revenue is projected to reach $3.7 billion this year. To achieve its ambitious goal of $100 billion in annual revenue within five years, Open AI must focus on finding valuable paid opportunities in both the C-end subscription market and the B-end enterprise and individual developer segments to attract users to pay for its services.

Since last year, Sam Altman's strategy of focusing on products has become evident.

Last November, at the first annual Developer Conference hosted by Open AI, the company not only unveiled the large model GPT-4-Turbo, which significantly reduced token prices while maintaining performance, offering a discount equivalent to 1/3 of the input and 1/2 of the output tokens compared to GPT-4, but also showcased their new product – GPT Store. GPT Store allows anyone to create customized versions of ChatGPT by setting prompts based on model capabilities and their own needs, without the need for coding.

GPT Store Source: Open AI

In July of this year, Open AI officially launched its AI search product, Search GPT, but limited the beta testing to just 10,000 users. "We believe there is still significant room for improvement in search today," commented Sam Altman on X on the day of SearchGPT's launch.

While experimenting with AI products, Open AI is also recruiting experienced product developers to find the right talent for its productization journey.

In June of this year, Kevin Weil officially joined Open AI as Chief Product Officer. According to the official announcement, Kevin will lead a product team. Prior to this, he served as Vice President of Product at Facebook Novi, Instagram, and Twitter, bringing extensive experience in product development.

However, unlike Open AI's absolute advantage in developing multimodal large models, its moves in launching new products, from GPTs to AI search to Canvas, have always lagged behind.

To date, every AI product released by Open AI has a comparable counterpart. For instance, GPT Store is not a novel feature. As early as the beginning of last year, several platforms that integrated the capabilities of large models from Open AI, Google, Anthropic, and others, and offered customized agent functionality, had already launched, such as POE, an AI bot alternative created by the CEO of Quora, the "overseas version of Zhihu."

As the debate over whether AI search can kill traditional search engines intensifies, Open AI's product launches have lagged behind both traditional giants and startups in the same space. In terms of product form and interaction experience, SearchGPT has not offered any groundbreaking improvements.

Similarly, the recently launched Canvas is not an innovative product either, and is seen as a competitor to Anthropic's June releases of Artifacts and Cursor, a programming software.

Artifacts, a product for code visualization Source: Anthropic

Open AI's ambition to dominate the implementation of AI application products is clear, but its moves have not been impressive.

Real product or fake demand?

Despite its continuous activity, Open AI's products released in the past two years have struggled to replicate the explosive virality of ChatGPT in the mass market.

According to foreign media reports, the launch of ChatGPT was a bold move that caught Open AI's board of directors by surprise. With ChatGPT achieving over a million users in five days and over 100 million users in 60 days, Sam Altman tasted the immense wealth generated by a flagship product.

Since then, while releasing its latest models, Open AI has also observed the practical application of these models in products. In these potential markets that could satisfy demand and generate new revenue streams, Open AI does not want to miss out, as evidenced by GPT Store and AI search.

However, after ChatGPT, Open AI's subsequent offerings have struggled to satisfy users again.

In nearly a year, GPT Store has failed to prove its potential as "the next App Store" or even retain its limited number of early adopters.

Like App Store, GPT Store's ecosystem relies on both users and developers. However, GPT Store's design is not developer-friendly, performing poorly in areas such as traffic generation and plagiarism control. For instance, it allows multiple users to create GPTs with the same name and does not restrict GPTs with high similarity. Without the ability to generate revenue or gain effective exposure, developers' creative enthusiasm is inevitably dampened.

Worse still, GPT Store fails to engage users. According to The Information, a developer studied over 30,000 GPTs and found that most of them had only 1-2 users per day, with only 5% attracting 150-500 users.

Similarly, large-scale testing of AI search remains pending, joining the video model Sora, which was launched earlier this year, as an unfulfilled promise from Open AI.

In terms of product form, Open AI's SearchGPT still focuses on the "answer engine" approach for AI-generated content, synthesizing vast amounts of internet information to answer user queries. Given the low barriers to entry in AI search products, Open AI has yet to introduce a more differentiated offering.

SearchGPT interaction interface display Source: Open AI

Even if launched, it remains questionable whether SearchGPT will truly steal the spotlight from other AI search products. Currently, from a revenue perspective, Open AI's cash cow still comes from subscription services and collaborations with enterprises and individual developers. These products have neither generated the expected traffic for Open AI nor supported a significant new growth curve in revenue.

However, in the coming years, Open AI is destined to expand its ChatGPT services. In documents disclosed by The New York Times, Open AI plans to increase the subscription price of ChatGPT from $20 to $22 this year, with the price doubling to $44 in five years.

Fortunately, the release of Canvas showcases Open AI's progress in product design and interaction. Whether users will be willing to pay for it remains to be seen by the market.

How to persuade users to willingly pay for expensive pricing? Canvas is just a small part of the value-added services. Finding more paid opportunities is a question that Open AI needs to answer with actions in the coming years.