Huawei's Pangu 3.0 launch (https://bit.ly/3D1Og9h) ought to serve as a shining example for large corporations, irrespective of tech or not, and governments across Asia. But alas, many in the non-US corporate realm fail to grasp the ongoing revolution, falling into one of two categories: the "We also do AI" chanters and the "We cannot do AI" chanters. Let's affectionately refer to them as the defensives and the defeatists for brevity.

The defensives fail to comprehend the profound impact and validation of transformer models through ChatGPT3.5 (https://bit.ly/45QZrhZ). While US tech companies immediately recognized the potential of exhilarating innovation paths ahead and effected substantial changes to their plans, the defensives remain stubbornly fixed on their predetermined paths. It's like the major phone makers when the iPhone burst onto the scene in June 2007, desperately clinging to their existing product plans and ultimately meeting their demise.

The defensives cannot fathom the transformation since the days before GPT3.5, mainly entangled in the bewildering maze of definitions surrounding artificial intelligence (https://bit.ly/3qdqYKi). As a result, they get stuck to likely outdated plans embarked upon ages ago. They lack the urgency to reconvene at the drawing board and forge new plans. Perhaps they should take a leaf out of Adobe, Salesforce, Huawei, or even the books of Microsoft and Google to truly comprehend the need for change.

The defeatists adamantly believe that building foundation models is an insurmountable task for them. They feel they lack the time, the budget, the right talent, or the data even to contemplate a plan (it must be said that this is likely true for many, but not all, and most notably, the largest). The short-termism involved in resigning to someone else's API could cost many dearly in the long run, not just in terms of fees or import bills at the national level but also in their ability to innovate or protect their proprietary data.

Building foundation models isn't rocket science (as argued at https://bit.ly/46cRGDh). If Huawei can unveil a model with five foundation models at its core, despite the challenges it must have faced in procuring the right hardware, then indeed, any sizable company or conglomerate from India, Korea, Japan, Singapore, or everywhere else stands a chance at developing their own narrow-AI models of the highest caliber.

The days when non-US/China or non-tech companies proudly announce their own models and groundbreaking paths - rather than mere partnership-based progress - are not far away, but they could be sooner.