Your Company: What Happens Tomorrow if OpenAI Collapses?

An increasing number of small and medium-sized enterprises (SMEs) are integrating ChatGPT into their operations. From handling inquiries and creating meeting minutes to drafting proposals and generating code, it can perform the work of more than one employee for just a few thousand yen a month. There’s no reason not to adopt it.

But I want to ask one thing: If OpenAI were to stop its services tomorrow, would your company still be able to operate?

You might think, “That’s impossible.” However, understanding what is currently happening within OpenAI should shake that optimism.

What is Happening Inside OpenAI

Reports have revealed a growing distrust within OpenAI from late 2024 to 2025.

Whistleblowers have publicly expressed their distrust towards CEO Sam Altman. While there are multiple points of contention, the core issue is that “the vision Altman promotes is disconnected from the reality of the organization.” Key members of the safety research team have been resigning, and lawsuits surrounding the transition from a nonprofit to a for-profit organization have emerged.

The recent turmoil surrounding Altman’s dismissal and reinstatement in November 2023 is still fresh in memory, but that chaos has not subsided; it has merely gone underground. The internal rifts are deepening.

This implies that the stability of the OpenAI platform is not as certain as it appears.

A New Risk: The AI Profit Tax

Another development that SMEs should pay attention to is the “AI Profit Tax” proposed by OpenAI itself.

This tax system aims to redistribute profits generated by AI back to society, addressing the concentration of AI technology benefits among a few large corporations. While the concept is understandable, the problem lies in “who” will bear the burden and “how much” they will pay.

Currently, the specifics of the system design are unclear, but if companies using AI services are taxed, the impact on SMEs could be significant. For instance, if a 5-10% tax is added to the cost of using AI tools, a monthly API fee of 20,000 yen would rise to 22,000 yen. That’s an annual increase of 24,000 yen. While this may seem small for one company, it can accumulate if multiple AI services are utilized.

More importantly, no one knows which country will implement this tax, when, or in what form, creating uncertainty that could suddenly change the assumptions underlying business plans.

The Implications of Overheated VC Investment

The AI industry is seeing a massive influx of venture capital funding. OpenAI’s valuation has surpassed $200 billion, and related VC funds are being established one after another, making $100 million fundraising rounds “normal.”

However, looking back at history, areas that attract huge amounts of capital inevitably experience bubbles, and those bubbles always burst. The dot-com bubble of 2000 and the cryptocurrency bubble of 2022 serve as reminders. While the influx of funds is generally positive, if that funding is based on “expectations” rather than a “sustainable business model,” adjustments are unavoidable.

When the AI bubble bursts, what will happen? AI startups struggling with cash flow will cease operations. API fees will skyrocket. Features will be scaled back. Free plans will be eliminated.

There is no guarantee that the AI services your company relies on will be excluded from this scenario.

Three Things SMEs Should Prepare For

1. Diversify Platforms: Don’t Put All Your Eggs in One Basket

The most crucial measure is to reduce dependence on a specific AI platform.

Specifically:

• If you primarily use ChatGPT (OpenAI), verify whether similar tasks can be performed with Claude (Anthropic) or Gemini (Google).
• If you are using API integrations, understand how many days it would take to switch to another API.
• Test once a month whether critical business processes can operate without a specific AI.

This acts as “insurance.” While insurance incurs costs, it’s too late once an accident occurs. The time required for platform diversification checks is half a day to a day per service. Doing this twice a year is sufficient.

2. Have Options for Local LLMs

As of 2025, the performance of open-source LLMs (such as Llama, Mistral, and Gemma) is rapidly improving. While it may not be realistic to rely entirely on local LLMs for all operations, creating a state where “at the very least, minimal operations can be handled locally” holds significant value.

For example, summarizing internal documents or creating standard emails can be adequately managed by local LLMs. The required hardware is just one PC with 32GB of memory, with an additional investment of around 100,000 to 150,000 yen.

3. Make “AI Expenses” a Separate Budget Item

Many companies blend AI-related expenditures into “IT costs” or “miscellaneous expenses.” This obscures both the degree of dependence on AI and the risk of cost increases.

Separate AI-related expenditures into an independent budget item and track them monthly. This includes API usage fees, subscription costs, and related cloud expenses. By consolidating everything into one category, it becomes immediately clear how much your company is spending on AI and what percentage of revenue that amount represents.

As a guideline, if AI-related expenses exceed 3% of revenue, it indicates too high a level of dependence. For a company with annual sales of 50 million yen, that would mean 1.5 million yen. If you exceed this threshold, a review of your cost structure is necessary.

The Line Between “Dependency” and “Utilization”

I’m not saying you shouldn’t use AI. In fact, you should actively use it. However, “using” and “being dependent on” are different.

Using: Business can operate without AI. AI makes processes faster and improves quality.

Being dependent: Business stops without AI. There are no alternatives if AI fees increase.

The former is a sound business decision; the latter is a failure in risk management.

Whether the internal collapse of OpenAI becomes a reality is uncertain. When the AI profit tax will be implemented is also unclear. However, preparing for “what if” scenarios is something you can start doing today.

Future Points of Interest

From late 2025 onward, the following developments should be closely monitored:

• The outcome of OpenAI’s organizational restructuring: Timing for the completion of its for-profit transition and associated changes in pricing structures.
• Discussions on the AI profit tax in various countries: Movements in the EU, the United States, and Japan.
• Performance improvements of open-source LLMs: When GPT-4 level open-source models will emerge.
• The culling of AI startups: Consolidation due to changes in the fundraising environment.

These developments will directly impact SMEs’ AI strategies. I strongly recommend establishing a habit of reviewing your company’s AI dependency every six months.

AI is a tool. If you become a user of the tool rather than the one using it, you have already lost.