Who wins the AI race? VAST Data’s CEO weighs in

Image created by DALL·E 3.

When it comes to AI, enterprises are accelerating their efforts to build the next big thing. However, it’s not always the organisations with the largest budgets or shiny new toys that reach the pinnacle of success.

From chatbots to ChatGPT, AI has inspired entire industries to devise innovative solutions to longstanding problems. With everyone speeding on the AI expressway, who will ultimately emerge victorious?

During a media briefing, Renen Hallak, Co-Founder and CEO of data infrastructure company VAST Data, shared his perspectives on the current state of the AI race, including ways to broaden the technology’s accessibility.

Opportune start

According to Hallak, the company’s inception in 2016 was impeccably timed.

“We’ve been lucky to start late. That’s a big advantage that we have over everyone else in this space. The fact that we were founded in 2016 allowed us to see the future. It allowed us to partner with organisations that have been on the bleeding edge of AI for the last few years, and they’ve been showing us what we need to build for them to be successful,” he said.

The Chief Executive further described VAST Data’s role as the software infrastructure layer for this era as “unique.” He also noted that predicting which application vendors will be successful is very difficult.

Renen Hallak, Co-Founder and Chief Executive Officer, VAST Data. Image courtesy of VAST Data.

To illustrate his point, Hallak drew an analogy with Cisco’s role during the internet’s rapid expansion. He observed that, much like the early days of the internet, it was difficult to foresee which websites would succeed, yet to him it was evident that Cisco would prosper by laying the foundational infrastructure for the internet revolution.

AI has evolved from a scary, larger-than-life concept in movies to a practical tool that enhances efficiency and problem-solving capabilities. Routine and repetitive tasks can now be automated, saving companies countless hours of precious time.

“The landscape is shifting from human-driven processes to machine-driven ones. For example, in image analysis, what once required a person to review each image can now be processed in real time as feeds come into a system through computer-aided visual capabilities. This allows for the immediate identification of individuals, activities, and anomalies, issuing alerts for any discrepancies. These advancements, previously unattainable, are now made possible by AI,” Hallak stated.

Competitive edge

Previously, a lack of data posed a significant challenge for AI. Today, while data availability has increased, storage capacity remains a limitation. Consequently, companies are significantly investing in physical infrastructure to develop use cases.

In 2023, the main problem for companies was the shortage of GPUs in the market. Hallak remarked that although the situation is now improving, a new and more significant challenge is emerging.

“Over the next two to three years, power will be the limiting factor. We are seeing data centres being spun up close to power plants, and new power plants being built. It’s no longer just about real estate or GPUs,” he explained.

To address the storage limitations of data warehouses, enterprises have turned to using data lakes, which allow for storing raw data without the need for structured formatting. Despite this, the growing demands of AI mean that even these data lakes are still not enough to handle its requirements.

This challenge is one that VAST Data aims to overcome, according to Hallak.

“We serve many big banks and hedge funds. For example, a hedge fund analysing trade data might previously have only reviewed data spanning one or two weeks. Now, with VAST, they can review up to 20 years of data. This extensive access isn’t limited to trade data; it includes Twitter feeds, news feeds, natural language, unstructured data, and satellite imagery. This vast range of information allows them to significantly enhance their trading algorithms,” Hallak stated.

Other sectors growing in usage of VAST’s platform include government institutions and life sciences.

“We see a lot of information — whether it’s from research centres, universities, or hospital networks engaged in genomics research and medical imaging — everything is now processed and enhanced by AI. We have a few companies that are doing AI-based radiology and developing new ways to analyse proteins and subatomic particles. These are the advances that this new technology can bring,” Hallak revealed.

VAST Data is also involved with customers in manufacturing who are developing digital twins. This allows them to simulate and test before actual construction in the physical world.

The company’s technology is also being used by makers of autonomous vehicles. With data as the driving force, these vehicles represent a shift towards a data-driven business model, a transition Hallak sees as becoming more prevalent.

“What we expect in the next few years, as AI capabilities become more apparent, is that more companies will shift to that business model. Consider how Google operates; it doesn’t sell products but instead monetises the information it collects. Why shouldn’t a car company do the same? Provide the car for free and profit from the data generated. We’re observing an increasing trend in this direction,” he said.

Regulation vs innovation

With governments around the world establishing the ground rules for AI, questions arise about how tech companies and businesses should proceed with innovation.

For Hallak, the new rules, such as the EU’s AI Act, should actually empower organisations in realising their AI projects.

“Governments all around the world are embracing this new development. Naturally, every new technology should be regulated considering it can do a lot of good, but there’s also the potential for misuse. We need governments to protect us against, or at least mitigate, the potential misuse of AI. From what I’m seeing, regulation is structured in a way that encourages adoption. In fact, governments are keen to see who can best leverage these capabilities,” he noted.

To counteract the malicious and unethical use of AI, laws like the EU’s AI Act, and others in regions such as ASEAN, include provisions for cybersecurity. According to the VAST Data Chief Executive, cybersecurity should always be baked into a company’s technology strategy, and not just be an afterthought.

“There’s a lot that went into ensuring our platform is secure. As new capabilities emerge, we will need new security functionalities, similar to how the internet developed. A few years after its inception, cybersecurity evolved alongside it to safeguard us from potential threats. The same will happen with AI: a new subfield of AI security will emerge, addressing issues like copyright infringement, data privacy, and, as these systems become increasingly powerful, ways to limit and ensure they don’t perform unintended actions. All of this should be part of the thought process as we venture into this new era,” he explained.

So who wins the AI race in the end? As of now, it’s too early to tell, Hallak admitted.

“I don’t know if OpenAI will win, but they’re definitely ahead of everybody else. On the infrastructure layer, and on the chip side, and definitely on the hardware side, NVIDIA is well ahead of everyone else. We are positioned right in between those two, providing the software layer that bridges the application and the hardware,” he said.

“It’s hard to tell who will win because we’re just at the very beginning of this race. I anticipate many twists and turns over the next few years and decades. But whoever figures out how to harness machines to solve human problems around disease, climate, and energy, that’s the end goal we’re all striving towards. Once that is achieved, it will benefit everyone. So in a sense, it doesn’t matter who wins so long as the technology is made available to everyone,” the CEO concluded.