Is AI sustainable for businesses?

I sometimes have some extra time on my hands on the weekend, which allows me to arrange my thoughts and write what I think could happen within the AI game.

A good few things have been happening in the last couple of days, from rumours of Apple investing in OpenAI to Nvidia using SMC for cost reduction.

LVMS: The New Frontier

OpenAI is pretty much losing the LVM game, which means Sora needs to be launched to the general public within the next couple of weeks or end up as a non-interesting video generator, where there is a relatively worthy competitor to it such as Kling.

Sora was first announced back in February, 7 months ago (6 months + 3 weeks if you'd like to be nitpicky about it).

Since the announcement of Sora being offered in a closed beta to select users back in February, Kuaishou launched Kling AI to the general public as part of its beta 6 weeks ago. This means that Kling AI is already making headlines and is used extensively, while Sora remains in the hands of whomever for an unknown period.

As such, OpenAI is losing the LVM game completely, whether to Runway, Pika and others or Kuaishou in that respect.

What Can OpenAI Do?

The most optimal option for OpenAI today would be to launch Sora within ChatGPT, possibly offering a third package for users:

1. ChatGPT Pro (which includes Dall-e 3 and Whisper already, as well as data analysis capabilities).
2. ChatGPT Premium (which includes Sora and SearchGPT as well as the other options in Pro).
3. ChatGPT Enterprise which includes all of the above.

OpenAI's Pricing Might Not Be A Good Fit For 2025

In terms of pricing, $20/month may not be sustainable enough in 2025, so either Pro remains at $20/month while Premium ups the ante to $35/month, rendering the reason to not upgrade and receive such powerful solutions at the juvenile level, or changes pricing and offer better inference within Pro to keep Pro users satisfied enough to stay within that sphere, maybe at a 10%-15% price increase.

A 10% increase could, understandably, lead to a higher switching rate (churn rate), especially if users already pay for a Pro version at MidJourney, which is undoubtedly the best image generator out there, or for Claude Sonnet which offers better reasoning and code writing than ChatGPT.

Even GPT-4-based tools, such as Julius, provide better options in terms of data analysis than the regular ChatGPT, putting OpenAI in a pickle with the Pro version.

Since ChatGPT's actual churn rate exceeds 20%, this might lead to a 25% churn rate on average, possibly greater.

As such, prudence in pricing and temperance in strategic expansion plans should be exercised to avoid reaching such numbers.

A Hodgepodge of Failures Post-Optimisation

ChatGPT Enterprise would probably be revamped with the failed integration of CoPilot into Office.

Though a great option at first, CoPilot 365 is a white elephant.

The cost of integrating it and learning how to work with it has rendered it a "nice to have" feature instead of a "must have" feature.

OpenAI though could present an invaluable offer altogether for enterprise clients in this respect: a downloadable version of ChatGPT directly into their servers, which would allow for better monitoring of data and the partial prevention of data leakage. This could either link up to Office 365 or even to older versions of installed Office, which would make the price of integrating it much better for smaller companies.

Can smaller businesses benefit?

If ChatGPT Enterprise though does come to be, may force Groq (which allows for the cheaper use of GPT 4, 4o and 3.5 as well as Llama and Mixtral) to offer specific offerings to enterprise clients, perhaps combining certain elements to keep afloat, as such an option does seem very much lucrative for enterprise clients who want to protect their data, without investing $4M-$5M in the process.

Such costs are not something most companies can endure, which may give larger enterprises the upper hand, even in an AI-driven world where costs are decreasing due to economies of scale.

However, with a more robust solution and perhaps if Groq goes to business with SMC (an ally of Nvidia), pricing will drop even more so, which would allow smaller businesses to integrate AI with ease, protecting their information from data leakage and at a more affordable price. Possibly not at the $500K level, but maybe within the reach of $1.5M-$2M.

This gives Agreggators as a Service (AaaS) companies (inference engines) the upper hand in the future of AI, providing a wholesale market solution for most companies.

Something to think about indeed!

Taking AI For A Spin! ?????

Are you ready to revolutionise your business processes with AI and unlock unprecedented efficiency and growth?

Contact me to explore how AI can transform your business landscape and discuss tailored AI strategies that align with your unique business goals.

#chatgpt #kling #sora #groq