As the CEO of NVIDIA stood on the stage with his latest technology marvel in his hands, the Blackwell chip, I was scared.

For someone who really loves technology, and finds artistic level beauty in technological applications and progress, you may ask, why was I scared? Was that not the time to celebrate?

When it comes to the exponential growth of the hardware power to support our AI dreams, this was (or is), huge. Yes !

The reason I get scared is something I have consistently highlighted before. Hardware capabilities now far exceed our AI dreams for the near-term. The challenge is, we have not been able to scratch even the surface with capabilities we already had, before Mr. Huang walked on the stage with this latest marvel.

To understand my concern, you have to understand that current AI capabilities can be demarcated broadly into two different areas.

One is the area where AI research by leading tech companies keep churning out more and more powerful versions of their flagship AI solutions. In terms of what they can do, the list is endless. And that list keeps getting longer. With Blackwell and similar products, that list will explode.

The other area is what has actually been done. And that is the worrisome area. The gap between what capabilities we already have, and what we have applied in businesses and society, is gargantuan. And that chasm keeps widening everyday.

With more and more powerful hardware capabilities, big tech can infuse more power in their solutions. With that, comes the press around how powerful these tools are now and what they can do.

The problem is, they are not doing what they can do. Not yet !

That means that the hype, which is true, since capabilities do exist, is growing significantly whereas real world deliverables are failing to catch-up, lagging far behind. And when this happens, there is always the possibility of bad things happening.

As capabilities of what AI can do, and the narration of the same, there is excitement, interest and FOMO. This can be channelized two different ways. One is to understand where we are, why capabilities have not been translated into solutions and tools, define a roadmap, accelerate the journey, so that the real-world deliverables match the capabilities.

The other way, where I see many going is where FOMO gets channelized into jumping on anything that breathes, is warm and has AI labeled on it. This is the where the true hype gets “gaslighted” into false expectations and hope. The result is, something that actually is capable of doing those marvelous things, may end up being labeled as false hype.

As I read about the HP Autonomy trial, it had similar tones. Back in 2011, analytics and real-time insights would have been the craze. Now think about it this way- we are in 2024 and I challenge you to find 20 companies that leverage real-time insights across the end-to-end enterprise (not one department, unit or division). If the charges are true, the hype was channelized into gaslighting, where the founders were able to inflate a product, matching that with the capability that existed. But the fact was, just like we have now, the gap between capability available (or what technology at that time could do) and what the product actually delivered, were wide. Just like today.

Every morning, I do an exercise with my morning cup of tea (yes, no coffee, blame the Britishers). I essentially go through my LinkedIn feed which is awash with posts and news about “AI-enabled” or “AI” companies. I research at least five of them everyday, just out of curiosity and habit. You know what the scary part is? In one entire week, I hardly come across more than one or two companies that meet the realms of true AI capabilities that exist today.

But VCs are funneling money into anything and everything, making things worse. Because if a product that actually never was capable of delivering what it could, or was not even true AI, fails, the interpretation is that AI is not delivering the promise.

So yes, I was scared when I saw Blackwell in Mr. Huang’s hands.

Because the chasm widens !