The Key to AI Deployment: Optimization through Infrastructure

While Star Wars became a household name during the 1980s, Intel remained a best kept secret.

That changed in the following decade when the company was the first to do “ingredient branding” – where one component of a product is the marketing star. The Intel Inside™ campaign created awareness as well as subsidies for OEMs lining up to display the iconic Intel oval swoosh on their computers. If you’re of age to remember, you probably still can sing the jingle: Doooo da-da-da-DAH.

Historically speaking, Intel Corporation has not been considered a strategic discussion point for chief information officers (CIOs) or chief technology officers (CTOs). Once again, however, Intel is changing the narrative. Through Intel’s alliance with Deloitte, the focus is on optimizing AI workloads and performance using Intel’s hardware as Deloitte puts infrastructure in the spotlight.

Infrastructure as a Paradigm Shift

The Era of Generative AI (GenAI) – not even two years old as of this writing – is quickly gaining traction. International Data Corporation (IDC) predicts investment in GenAI to reach $143B by 2027, up from $16B in 2023. With AI comes a paradigm shift reframing infrastructure as a strategic decision with 3 essential U’s: central processing unit (CPU), graphics processing unit (GPU) and microprocessor unit (MPU). Why? We live in a more connected world so as AI and the internet of things (IoT) converge, massive amounts of data demand more memory, more energy and more compute power. With each generation of Intel Xeon, for example, CPUs are becoming more efficient. This trend is a must-have for AI deployment.

According to Deloitte’s “Now Decides Next: Getting Real About Generative AI,” the second in a series of quarterly global reports on how businesses are approaching GenAI, many respondents are moving past “experimentation, pilots and proofs of concept” to scaling deployments with 60% reporting they are “effectively balancing rapid implementation with risk management.”

The report cites four top priorities for organizations: value creation, scaling up (expanding a GenAI’s user base), trust and evolving the workforce. In particular, scaling up and trust are the twin pillars that will support the widespread adoption of GenAI.

If you’ve ever studied piano, you know practicing scales impacts results. While subtle, Deloitte’s report reveals something similar. “Expert” organizations in GenAI are future-forward thinking, meaning they are diligently preparing for the future, building their agility muscle and leveraging GenAI to architect future growth opportunities. They are comfortable not having all the answers. For them, it’s about progress, not perfection.

• Of the 45% of organizations planning to reinvest savings from GenAI into innovation, those with “very high” GenAI expertise are driving innovation (51%) with an emphasis on new products and services rather than reinvesting savings in improving operations.

• Organizations with very “high” GenAI expertise are deploying more rapidly – 73% versus 40% of those with “some” level of expertise.

• Those with a “high” expertise invest in cloud, and use code generators and open-source LLMs more than those with “some” level of expertise.

The challenge, however, remains the same for organizations with or without a high level of expertise: which infrastructure, hardware and software do you choose? What model size and data type? Who gets AI PCs across the organization? What’s our security posture? What do I use for cloud computing, edge and data management?

Wouldn’t it be nice if someone could just hand you a playbook with all the answers? Deloitte and Intel did exactly that.

Simplifying the Question: Which Infrastructure Components?

Deloitte and Intel Corporation have developed a fit-for-purpose hardware and software framework and methodology to optimize AI workloads. Since consulting is in my DNA, I know simplifying parameters and prioritizing expectations by tech user gets you closer to solutions based on software requirements. It takes the headache away from making those decisions. Just a few of the parameters they address include model size, data type, frequency, latency and token size.

I wish I’d had such a framework six years ago when I first deployed AI for a client. Selecting the right data set, training the model and making the myriad of choices was challenging. Challenging, perhaps, is too convenient a word. Better adjectives might be tedious, frustrating and complex. Fast forward to 2021. Practically overnight, ChatGPT gave everyone access to GenAI without having to be a data scientist or coder.

No one person or company possesses all requisite capabilities to architect AI. The journey demands a collaborative ecosystem where diverse talents and resources converge.  This is why, in the Era of On Demand, Intel is more relevant today than ever before. Enterprises crave more than mere efficiency; they seek performance, energy and cost efficiency at scale.

What’s inside a computer has become a strategic and environmental decision for top executives. The new Intel processor rises to the occasion.

Unveiling the Potential of AI Infrastructure

My new Microsoft AI PC with Intel Core™ Ultra processor is 13 inches, the same size as my Mac, but is different in every other way. Battery life is 19 hours which means that I don’t sweat if I leave my power cord behind. A touch screen allows me to whiteboard ideas and get to files faster. Microsoft’s Copilot, called an “AI Companion,” assists me with tasks like note taking during meetings and email organization. The AI camera elevates my conference call experience.

Yet, the full AI experience would be impossible without what’s inside the laptop: Intel’s processor. The efficiencies from Intel’s new chip and its space and energy savings allows us to do so much more with less.

In an interview on my show CXO Spice with Shakir Rizvi, senior manager business strategy and Intel alliance leader at Deloitte, we explore AI optimization. His passion for emerging technologies is unmistakable. The time is now. He points out 94% of participants in Deloitte’s report state that AI is critical to their success in the next five years.

While attending Intel Vision 2024 early April in Phoenix, Intel CEO Pat Gelsinger said, “Intel will unleash the power of our data, accelerate our productivity, enable our workforce, vastly improve the ROI of our technology investment, all while making it sustainable and secure.” This was before he danced onstage while holding the latest chip. It was a historic moment. We knew, each of us, that we were witnessing something big.

Pat was a dozen years into his first career at Intel when the Intel Inside campaign launched. He rejoined the company as CEO in 2021, picking up where the familiar da-da-da-DAH tune left off. Today, Intel and Deloitte are writing a new score where infrastructure once again takes the stage.

As we embrace the era where infrastructure becomes the linchpin of AI innovation, the journey forward promises not just efficiency, but unprecedented possibilities for growth, productivity, and sustainability. With infrastructure as the cornerstone, the path to realizing AI's transformative power is clearer than ever before.