From ChatGPT to AGI: OpenAI's Pricing Strategy, Enterprise Expansion, and the Path to Achieving AGI
In recent years, OpenAI has consistently remained a focal point due to the rapid development of AI technology and its societal impact. In particular, an interview with OpenAI's Chief Operating Officer (COO) provided specific insights into the organization's culture since its early days, the challenges and vision associated with global expansion, and future product strategies and initiatives toward AGI (Artificial General Intelligence). Based on the content of the Wall Street Journal interview, this article provides a multifaceted explanation of the mindset with which OpenAI is advancing its technology development and business expansion, as well as the outlook for future device strategies and enterprise-focused business.
1. YC-style Venture Mindset and OpenAI's DNA
OpenAI's COO first revealed that both they and CEO Sam Altman are alumni of Y Combinator (YC), stating the following.
"Actually, I hired Sam at YC (laughs)."
Starting with this humorous exchange, they emphasized that the YC-style "venture spirit" is deeply rooted in OpenAI's foundation. Specifically, the following elements were mentioned.
Experimental Approach
OpenAI emphasizes a research mindset, advancing diverse projects in parallel "like letting many flowers bloom."
"We must think from the perspective of what kind of results we can achieve if this succeeds."
Founding as a Research Project at YC
The interview also included the statement, "OpenAI originally started as a research project at YC," reflecting an organizational culture that fuses entrepreneurship with a research-oriented stance.
"We have unconstrained thinking and incorporate a venture mindset as much as possible."
Platform Orientation
Sharing the idea that, like YouTube or applications, AI is "a new platform for the Software 2.0 era," from which much innovation will emerge.
"If YC embodied many of the things that should be built with Software 2.0, then OpenAI is the platform that should be built in the AI era."
In this way, OpenAI explores market opportunities without fearing risk, conducting experimental initiatives based on learning. It can be said that the "try quickly and learn from failure" attitude inherited from YC is the driving force supporting their current rapid growth.
2. Data Center Expansion in the UAE and the "Democratic AI" Concept
A major topic covered in the Wall Street Journal was OpenAI's participation in the "Stargate AI Infrastructure" project currently underway in Abu Dhabi. While it was reported that Elon Musk expressed dissatisfaction with this, the COO explained the background as follows.
OpenAI for Countries Initiative
A program that began in response to "country-level demand" for nations to develop AI infrastructure and improve their own competitiveness.
"Many countries want their own version of 'Stargate.' AI will become an extremely important element for a nation's economic growth and public services, so there is a strong need not to fall behind."
Reason for Choosing the UAE as the First Testing Ground
The UAE was judged to be the optimal first pilot because it is adept at infrastructure construction and is a country proactive in technology adoption.
"The UAE is a very technology-forward country, and it was ideal as an experimental environment to gain many insights."
Values of "Democratic AI"
The interview discussed "how to protect the infrastructure and values necessary to build democratic AI."
"The AI platform we provide and the technology built upon it must fundamentally support democratic values."
However, since the UAE is not a democratic nation, specific operational rules and governance, such as "whether citizens can freely access ChatGPT" and "handling censorship or speech restrictions," will be future challenges. The COO also stated, "We will consider those parts while cooperating with the UAE government and the US government," acknowledging that it remains in an adjustment phase.
3. Rapid Growth of ChatGPT and Enterprise Expansion
In the interview, the COO also mentions in detail the usage status of ChatGPT and the current state of the enterprise business.
Trends in user numbers
Recently, it has recorded 5 billion monthly visits and approximately 500 million weekly active users, continuing to grow at a breakneck pace as a consumer service.
"500 million people are using it every week." (500 million users a week is a crazy number)
Shift to enterprise
ChatGPT for business (ChatGPT Enterprise) and related products increased by 50% in three months from approximately 2 million seats to 3 million seats as of February 2025, which the COO described as "explosive demand."
"The growth curve for business is different from consumer, steady but very fast. We are starting to see diverse use cases such as education and healthcare."
Specific implementation examples
California State University system: Providing AI-powered curriculum design and personalized learning for 500,000 students, the COO expresses hope that it "could become the first AI-enabled university system in the U.S."
"We are collaborating on various uses such as curriculum planning, career preparation, and individualized learning."
Moderna (pharmaceutical company): Realizing AI utilization in drug discovery processes and research and development, "it is being used in completely different ways in the medical field as well."
"It's a completely different use case, but it's amazing that the same technology can power both."
Differences between enterprise and MS/Google suppliers
While Microsoft and Google compete with their own clouds in the background, the COO cites "the integration of model development and product development" as OpenAI's strength.
"We are the most active users of our own models, and we are deeply familiar with the model training process and design. We immediately grasp what it is good at and what it is bad at, and getting support feels like consulting someone right next to you."
4. Pricing Strategy and Revenue Model Experiments
The COO also speaks frankly about ChatGPT's pricing.
An "experimental price" of $200/month
While many consumer subscriptions are considered to have a "$20 cap," OpenAI raised ChatGPT Plus from $20 to $200 at the end of 2023.
"The $200 amount was something Sam and I decided on a whim, saying, 'We have to pick a number somewhere, isn't $20 fine?'"
The existence of hyper-users and costs
As a plan for "power users" who use ChatGPT for many hours every day, the goal was to create a structure that could pay for costs even at a high price, covering server operation costs and tuning costs.
"The usage distribution of ChatGPT is an extreme power law, where a very small number of users use a huge amount. It was difficult to support them unless it was around $200/month."
Possibility of future price ranges
When asked jokingly by the interviewer, "Is there a $2,000 plan?", the COO replied, "(Laughs) I hope you pretend you didn't ask that, but we will continue to experiment with price ranges," showing a stance of adjusting while watching market reactions without being fixated on setting an upper limit.
5. Roadmap to AGI and Model Development Cycle
The COO also shared insights regarding the achievement of AGI (Artificial General Intelligence), one of the ultimate goals set by OpenAI.
Prospects for 'Achieving AGI by the Second Trump Administration'
While noting that 'Sam Altman said we would achieve AGI by the second Trump administration, and it remains a possibility,' the COO also remarked on the difficulty of defining the timeline and the concept itself, stating, 'The definition of AGI is ambiguous to begin with; if you ask five researchers, you get seven different answers.'
“Systems with capabilities that could be called AGI may emerge in some form within the next four years. The cycle for the O-series (O-model) has been shortened from the 12–18 months of traditional large-scale pre-trained models (GPT-series) to a matter of months.”
The O-series and the Speed of Innovation
It was emphasized that while GPT-3 and GPT-4 required long cycles for large-scale pre-training, the O-series allows for significant performance improvements in a short period of 'a few months.'
“Beyond improvements in the weights and design of pre-trained models, context understanding and reasoning capabilities have improved dramatically, and we have entered a phase where we are approaching AGI at a faster cycle.”
'Post-AGI' Planning for Enterprises
Addressing the difficulty for CIOs in planning more than five years ahead, the COO advised, 'Unlike Cloud or Mobile, the themes discussed each year in the AGI era will change. Companies need to build AI understanding and organizational muscle early and update them continuously.'
“Even for a major financial institution like Morgan Stanley, it took two years to build the foundation for AI utilization. If you don't start investing now and root AI literacy within your company, you will be left behind in five years.”
6. Partnerships with Publishers and Content Utilization
Regarding the use of external content for AI model training and inference, collaboration with news media is receiving particular attention.
Partnerships for 'Reference' Rather Than 'Training'
The COO asserted, 'Even if you aggregate all the content from publishers around the world, it accounts for less than 0.1% of the total training data for GPT-4. Therefore, its value as training data is small,' explaining that the primary purpose is 'to give the model rich perspectives and color, and to use it for search and reference information.'
“We want to use news information to keep the model's knowledge up to date and use that to provide users with accurate references.”
Support for Local Publishing
Given the difficult industry structure of local media, the COO revealed an initiative to 'collaborate with local publishing companies to revitalize regional news using AI, as highly topical local information is the hardest for AI to grasp.' They are reportedly working with organizations like Axios to build a 'system that automatically collects and distributes regional information, such as for Topeka, Kansas.'
“When you want to plan your weekend leisure, the experience of knowing 'what is happening in the neighborhood' creates tremendous value. Supporting local media will continue to be an ongoing theme.”
Involvement with Hollywood
Although a research preview of the video generation AI 'Sora' has been released, the COO honestly admitted that it is 'not yet of a quality that can withstand production use for filmmaking.' However, multiple video creators are already using it experimentally, and the company has received feedback that it will eventually become an 'indispensable tool,' looking forward to expanding collaboration with the media industry as the technology matures.
“Currently, it is a research preview, but it is highly valued for enhancing creative workflows and is very useful in the ideation and initial concept exploration phases of video production.”
7. Potential of Agentic AI and Future Device Strategy
Toward the end of the interview, the phase where AI evolves from a mere chatbot to an 'agent,' and the device vision that supports it, were mentioned.
Evolution of Agentic AI (Operator)
At this stage, it is still just a challenging research preview, but the COO explains, 'Agentic AI goes one step further than providing knowledge via chat; it has the ability to actually operate tools and execute tasks.'
“Operator is something that allows AI to move a mouse and keyboard like a human, enabling web operations and application control. It is still in the experimental stage, but I am convinced it will become an essential technology in the future.”
Acquisition of IO, Led by Jony Ive, and Ambient Computing
The interview touched on the acquisition of the remaining shares of IO, the company led by legendary Apple designer Jony Ive, and the integration of design talent into OpenAI. The goal is to 'provide next-generation AI-equipped devices in an “ambient” (blending into the surroundings/environment) form.'
“Every time a new foundational computing platform emerges, corresponding devices appear. AI is no different; it requires a new set of devices that create an ‘ambient’ experience.”
“An AI that is personal yet harmonizes with the environment”
While current ChatGPT is provided as a smartphone app, it was pointed out that “people don’t necessarily keep their eyes on the screen during their busy daily lives.”
“In human conversation, the way we speak changes depending on the situation and the relationship with the other person. AI also needs to switch its conversational tone between when it is with family and when it is with colleagues. To achieve this, we need dedicated devices that understand the environment and context.”
Although the specific form factor and release date remain undisclosed, when asked, “Is it an orb-type or an Alexa-type?”, the response was “no comment,” while emphasizing that they are in an “experimental stage of exploring the optimal solution through multiple prototypes.”
8. ROI and Productivity Improvement Case Studies
Finally, the topic of “ROI (Return on Investment) from AI adoption,” which concerns many companies and investors, was also addressed.
Productivity improvement rates by industry
Developers (software engineers) are reportedly seeing a 50% to 2x increase in coding speed by collaborating with AI.
“Most startups say, ‘The majority of the code being written is either in collaboration with AI or written entirely by AI.’ This is astonishing.”
A similar trend is seen in data science and quantitative analysis, and in the creative field, “improved iteration speed through idea generation and creative support” is being valued.
Further acceleration expected as agentification progresses
At present, usage is centered on chat and tool integration, but the COO predicts that “in the future, agents that execute tasks across multiple software and data sets will emerge, enabling cross-organizational operational efficiency.”
“Right now, the focus is on supporting fragmented information processing, but once agents evolve to the stage of maximizing productivity, the quantitative ROI will increase even more dramatically.”
The difficulty of quantification and a proposal for a qualitative approach
However, it is frankly admitted that “because everyone uses AI differently, ranging from a 2% efficiency gain to a 10x acceleration, it is difficult to uniformly quantify ROI.” With that in mind, the advice given was, “First, experience ChatGPT yourself, find the ‘power users’ within your team to qualitatively understand how it is used and its effects, and share those stories.”
“I recommend that CIOs first introduce it on a pilot basis to raise AI literacy within the company, collect user case studies, and then roll it out fully.”
From this interview, a multifaceted vision emerges clearly: OpenAI’s “venture mindset,” “commitment to democratic AI,” “acceleration of enterprise deployment,” “ambition to reach AGI,” and “investment in next-generation devices.” Building on the YC-style culture cultivated since its founding, and while gaining insights through projects around the world, the company continues to pursue its grand mission: “democratize AI,” “provide advanced computing resources to everyone,” and “ultimately develop AGI that supports humanity.”
The full picture of the future OpenAI is painting—including the experimental initiatives in the UAE covered in this article, the “ambient computing” concept welcoming Jony Ive, and enterprise adoption cases and corporate partnerships—is still fragmented. However, what is common to each of these is the attitude of “always experimenting, learning, and updating,” which can be said to be the greatest driving force for continuing to lead the AI era. We look forward to following the progress of the new technologies and partnerships that OpenAI will continue to announce.
Related Articles
