Home Latest Insights | News From Intel to Nvidia, DeepSeek Provides Basis To ReThink AI Future on Speed and Computing Capacity

From Intel to Nvidia, DeepSeek Provides Basis To ReThink AI Future on Speed and Computing Capacity

From Intel to Nvidia, DeepSeek Provides Basis To ReThink AI Future on Speed and Computing Capacity

Focusing on “speed” as the core metric of performance can lead to some own-goals in the world of tech. Intel became a category-king and dominated its era when its unalloyed religiosity on sustaining Moore’s Law was the nucleus of its business model. But as it focused on speed, pushing down the feature sizes of transistors, it forgot the “besides speed, are you checking other metrics”.  From the flanks, new species of companies like Nvidia, Qualcomm, ARM, and AMD 2.0 emerged, and punted the dominance of Intel.

Yes, if your phone is fast, are you open to trade that speed if the phone battery can last longer? In electronics, that I^2R in CMOS transistors and broad power dissipation will define the usability of products, well ahead of the speed you can get things done. So, when those new ARM-anchored processors came out with better power efficiency, markets changed in the mobile internet era.

Then we moved into the AI fledgling era, and Nvidia, AMD 2.0, etc became the new Intel signing the songs of speed, processing capacity, etc as they make the case why high compute is the holy grail to unlock the heart of the AI age. But with that fixation on computer power, the crusaders like Microsoft, OpenAI and Google concluded that to win, more computing power is the denominator. Yes, efficiency and optimization were not really vital.

Register for Tekedia Mini-MBA edition 17 (June 9 – Sept 6, 2025) today for early bird discounts. Do annual for access to Blucera.com.

Tekedia AI in Business Masterclass opens registrations.

Join Tekedia Capital Syndicate and co-invest in great global startups.

Register to become a better CEO or Director with Tekedia CEO & Director Program.

But from the flank, a small Chinese company named DeepSeek went for that metric of optimizing everything to use less computing power. Magically, we’re learning again that speed, computing power cannot be the only core metrics if one expects the system to be symphonically architected like an orchestra that delivers nice music-of-AI-services.

In the industry, one of my core roles used to be creating the electronics for MEMS systems. In pacemakers, when you engineer the MEMS-based gyroscope and accelerometer, you have to be really great on power management so that the electronics can be implanted in a human and stay there for more than a decade! That disciplined approach of designing for usage where scarcity defines the operating environment is the mindset which must be used for this AI age.

Simply, from the $500 billion Stargate to building nuclear power plants, AI leaders must chill and say, “could there be another way”. DeepSeek has provided a reset to rethink the future of AI at the foundation model, large language model and generative AI levels.

Tech stocks recovered Tuesday, with the Nasdaq rising 2%, after China’s DeepSeek AI model sent shockwaves across the sector and cast doubt about U.S. dominance in the artificial intelligence race. Nvidia shares rebounded, climbing almost 9%, following Monday’s 17% rout, which erased a record $600 billion in market value. With U.S. tech giants reporting earnings this week, one analyst sees a “real potential shift in the underlying narrative” as investors assess the high valuations of many AI-linked stocks.


---

Register for Tekedia Mini-MBA (June 9 – Sept 6, 2025), and join Prof Ndubuisi Ekekwe and our global faculty; click here.

No posts to display

1 THOUGHT ON From Intel to Nvidia, DeepSeek Provides Basis To ReThink AI Future on Speed and Computing Capacity

  1. Francis Oguaju says: January 29, 2025 At 12:37 AM

    When the person holding both the knife and yam is telling you how expensive the yam you are waiting to eat is, if don’t have a way of challenging him, you end up agreeing that the yam is that expensive. What we heard for months from Nvidia and co was release after release of chips, with billions and trillions of transistors added. The ones with plenty influence were busy raising money and squandering, all in the name of faster compute.

    When are we going to learn that so many things are grossly overpriced? Is it just uncontrollable greed or ignorance? Of course prudence does not rank highly when spending other people’s money. We will know how long the deceit will last, until a new deceit comes to town. The play continues.

Post Comment

Please enter your comment!
Please enter your name here