The Strategic Review Applied to the AI Market

1 The Positioning

Following Kotler, depending on their strengths, tech firms must position themselves as

. a market disruptor

. a market challenger

. a market incumbent

. a market follower

1.1     Industry Analysis

The strategic review always begins with an industry analysis. In the IT industry, almost like clockwork, a new market disruptor emerges every ten years. Today, that disruptor is of course OpenAI.

1.2 Market Leader Strategy (in tech the Market Disruptor)

Here too, the strategic review does not begin with your firm's business strategy but with that of the market disruptor, namely OpenAI. It  consists of the following phases:

Initial Product 10 years on after its founding in 2015, OpenAI has successfully come up with a marketable product, comprised of  a program (the large-language model or LLM), a database covering everything publishable online (GPT), and a chatbot (ChatGPT). This first phase can be called the core product development or just product phase.

Go-To-Market Financing It successfully clinched already in 2019 $1B from Microsoft to finance its product development. Killing two birds with one stone, by 2023 it received another $12B from them to finance its commercial development which enabled it at the same time to tap into Microsoft's customer base of 1.6B users and 2M businesses. This second phase can be called the go-to-market product financing or money phase.

Standards As in any industry, the 3 to 5 major players (OpenAI, Anthropic, Meta, Google, DeepSeek) must agree on an industry-wide standard, which is the only way for individual firms to benefit from economies of scale and bring down costs. This is why OpenAI and Anthropic are pushing to have their LLM models, respectively o3 and MCP, adopted as the standard for AI, just as Linux became the standard for operating systems, and distribute it at first for free to lure the greatest number of users. This third phase can be called the scaled product development or industry phase.

Partnerships OpenAI has recently acquired for $6.5B Jony Ive's design firm iO, in order to fill a missing gap in hardware, and this time to lure the greatest number of buyers. This fourth phase can be called the differentiated product development or market phase.

Final Product User applications, called agents (hence the term "Agentic AI"), must now be scaled up. This fifth phase can be called the application development or monetization phase.

1.3 Market Challenger Strategy

A challenger to watch closely is Elon Musk's xAI. Just as Apple was also a challenger until it bypassed in a few short years Blackberry, Elon has  very quickly narrowed OpenAI's lead i) by releasing the right products in his Grok AI engine,  his AI chip for his Dojo supercomputer and Colossus data center, ii) by tapping into the right market of 1.6B users, iii) through the right channels at X and now at Telegram. Investors, who have put in $27B, have seen xAi's valuation increase 8-fold to $200B in a year.

1.4 Market Incumbent Strategy

OpenAI, backed by Microsoft, Anthropic backed byAmazon,  and xAI backed by Elon must face off the most formidable incumbent that is Google. The search giant has made great strides, having nearly finished making end-user tools which the 3 upstarts do not yet have. To go against OpenAI's AI engine ChatGPT, Anthropic's Claude, and xAI's Grok, Google must have easily matched the 3 upstarts' $10B to $15B of capital commitments, in order to develop and roll out its own AI engine, Google Gemini.

To understand the extent of Google's formidable reach, let us not forget that the founders of OpenAI, Anthropic and now Safe Superintelligence, Inc (SSI) cut their teeth in their twenties as crack developers at Google's in-house AI incubator, Google Brain (notably Ilya Sutskever, later chief scientist at OpenAI and after a falling out with Sam Altman now at his own firm, Safe Superintelligence, and the Amodei siblings at Anthropic).

However one wants to cut it, the Internet business we've now rebranded as AI is still Google's "sort and search" business extended now as a "sort, search, and solve" business. As of mid-2025, we can do search in "AI mode" on Google's website.

1.5 Market Follower Strategy

It is only after reviewing the competitive landscape and the competitive strategies of those who are in fact the market leaders (Open AI, xAi, Google,…) that one can begin to tackle one's own strategic  review as a market follower. As shown below, the firm has to find the right product/ market/channel fit. Using such tools as the 4 Ps and the 5-Forces, it must develop not one but 3 strategies, its business, competitive, and marketing strategies, by iteration.

The Chinese upstarts, not only DeepSeek but now also 01.AI, Minimax M1, MoonshotAI, Zhipu AI, Baichuan Intelligence and Stepfun, could be such market followers. They are backed by China's own FAANGs, Alibaba, Tencent, Baidu and the Chinese government.

2 The Business Model

They can now develop their business models. The map below, shows who the biggest players are, as reflected by the size of the circles (Open AI in purple, Google in yellow and the Chinese in orange), operating their AI data centers powered by their AI chips and AI databases and running their AI engines. So far, as their own chips are still under development, only Nvidia GPU chips are used.

Perhaps more importantly, the map also tells us what their distinct business models would be. To begin with, using the 4 Ps, in terms of the Product, they are tailoring and pricing their software, the AI engines, to meet the needs of their target markets. For example, Google has on the one hand their top line Vertex AI LLM, priced higher to process resource-intensive apps for those of their Fortune 500 customers needing not just text but images and videos and on the other hand, their Coden, Imagen, and Chirp LLMs, which are customized into more narrowly defined apps and priced lower.

In terms of the remaining 3 Ps, Price, Place and Promotion, nothing has changed. They will continue to use dual pricing, billing a fixed subscription fee and a variable usage fee, based on the number and complexity of the queries known as tokens. As for the marketing channel downstream, because they target the entire total addressable market or TAM, they will partner with the usual intermediaries, the wholesalers and the retailers. We will therefore continue to have a Kyndryl, an IBM spinoff, to operate the data centers under the AIOps moniker and IBM itself to provide the higher margin installation services using the watsonx and Granite portfolio of AI tools. Just as they did with the Cloud and the Web, these IT service providers would offer to "integrate" this time around the latest craze, AI, into our IT systems.

3 China

Finally there is the biggest threat of all, China. As we saw above with the top US market leaders, the 3 critical success factors are i) the product's software  (AI engine), and hardware (AI chip), ii) the market's billion+ users (AI apps using AI agents as middleware to adjust the application user interface or API), and iii) the money.China fulfills all 3 criteria but unlike us, without neither overvaluing nor overleveraging.

On the software side, the top Chinese Internet companies, such as Alibaba, Tencent, Baidu, and an OpenAI clone, DeepSeek, are already at the 3rd version of their AI engines, respectively Qwen 3, HPI, Ernie 4.5 and DeepSeek.

On the hardware side, Huawei holds 80% of the AI infrastructure market, equipping its Mindspore chip software to run its Ascend 910D (and by 2027 its 920 AI quantum computing chip) on its 245 TB OceanDisk EX 560 solid state hard disk.

On the user side, it is again Huawei who provides both the Harmony OS software and its Kirin 9020 chip, both of which power its family of Mate smartphones and its car systems for its co-branded EVs. Kingsoft's WPS Office, which runs not only on Harmony OS but also on Windows, OS, Android, and Linux, and which moreover is written in our Delphi/C/C++ programming languages, now competes head on against Microsoft Office. Should we not move to Harmony OS as well?

The Money Then there is the invisible hand of the Chinese government. As it has done with EVs, it has already shelled out this year alone some $60B of subsidies to the Chinese IT industry (our governments have also subsidized as much but without any strings attached, throwing good money after bad).

The Brainpower That the Trump Administration has pushed forward its $500B Stargate project with Oracle, OpenAI and Softbank and forced Apple to manufacture at home is a good sign. The question now is how the US will train 4 to 5 times more than the 140k engineers it currently trains to catch up to the 1.4M engineers the Chinese churn out every year.

4 DeepSeek

DeepSeek, Moore Threads, Humanoid Global. Have you heard of these firms? The first video below is a profile of DeepSeek's founder, Liang Wenfeng, a multi-talented entrepreneur who, after completing his master's in computer science in 2010, developed already back then an AI stock-trading program to take advantage of the capital markets' instability and volatility after the 2008 GFC. Capitalizing on its success, he created his own hedge fund, raising up to 19B yuan. In a sense, Liang was already stress testing what was to become DeepSeek. He applied it to the most difficult usecase there was, using the same type of AI algorithmic trading we had at home, to invest in the Chinese stock market for his clients.

It therefore came as no surprise that Liang  took a step further in 2019 by founding High Flyer AI to develop what was to become DeepSeek. When the first R1 version of DeepSeek was rolled out in January 2025, upon learning that its development had cost a mere $6M compared to the tens of billions of dollars raised by OpenAI, US tech stocks tumbled, losing $1T in market value.

The second video profiles Moore Threads, a Chinese start-up which was oversubscribed more than 4 000 times for up to more than the market cap of Nvidia or $4T. It makes GPUs to rival Nvidia. Then there is Humanoid Global, a Chinese VC fund with backing from Amazon, Google, and again looming large, the Chinese government, which invests in humonoid robotics, which can replace with "lights out" humans 24/7, 350 days a year non-stop.

5 The Future of Work

As the economic stats on China and the US show below, 90% of the jobs in China and the US are not in big corporations but in "low-tech" small and medium enterprises (SMEs). It is also these low-tech SMEs which have witnessed quasi uninterrupted revenue and job growth since the oil shocks of 1973 and 1979 and numerous recessions which followed. By contrast, most of the tens of millions of jobs lost since then, over the last 70 years, when the US transitioned itself out of heavy industry, have come from the big corporations, In fact, the buzz around AI has provided the ideal excuse for the "high-tech" giants (Meta, Amazon, Microsoft, Google, Intel, Accenture,...) to lay off tens of thousands of their workforce.

In spite of government policy favoring Big Business, these small, low-tech businesses have continued to show remarkably resilient entrepreneurship. These small businesses, with less than 500 employees, account for 40% of US GDP to the big corporations' 60%.

The real challenge ahead, which the Chinese have apparently understood judging by their 2024 economic report below, is to channel significantly more of our still enormous wealth into helping small businesses achieve stable and sustainable growth at home and abroad.

AI Disruptor

AI Disruptor

People

The Strategies

Customers

see Meta Case Study

Product, Market, Channel

see NFTs Case Study

How We Do It

Strategy

How We Conduct A Strategic Review

(download pdf)