Global AI Investment Boom Shifts Focus to Building Foundational Infrastructure

Today's massive investments in AI are funding a global race to build foundational infrastructure. As leading organizations build vertically integrated "AI factories," they're positioning themselves to own the technology that will power entire industries. According to some experts, the current gap between infrastructure spending and widespread consumer adoption is a predictable stage in a long-term build-out. Instead of asking "Are we in an AI bubble?" they say the more important question is: "Who will own and control the platforms that shape our future?"

For expert input, we spoke with Shahab Siddiqui, a Digital Strategy Consultant in the telecommunications and technology sectors across North America, the Middle East, and South Asia. With a career spanning senior sales, business development, and management roles, he has a proven track record of leading multi-million-dollar projects, driving sales growth, and implementing AI and automation solutions for government and enterprise clients. Today, Siddiqui identifies a fundamental mismatch in how technology is valued.

"You can’t put a price on operational resilience. The true return is long-term stability and competitive advantage," Siddiqui says. "Financial analysts who ask about the immediate return on investment are missing the point. Imagine Financial Analysts asking Telcos for immediate return on investment on expensive and expansive cellular network infrastructure in the late 90s when cell phones were not common."

While many see AI's commercial potential, its first high-stakes application is related to national security in the information space, Siddiqui continues. Here, the main challenge is widespread misinformation.

• Seeing is deceiving: Powerful and public generative AI is shifting the focus from securing servers to ensuring information integrity on a massive scale, Siddiqui explains. "The technology has become so convincing that even a fake video of the famous scientist Neil deGrasse Tyson required him to debunk it personally. The voice modulation and facial expressions were very convincing."

• Fact vs. fabrication: This new reality gives falsehoods a natural advantage, Siddiqui says. "This space is incredibly difficult to govern, especially when research shows that misinformation travels seven times faster than fact."

The technology’s accessibility means the challenge now lies in maintaining trust, Siddiqui continues. Instead of simply reacting, however, a more strategic response is to establish control of the core technology.

• Infrastructure as insulation: For example, Siddiqui points to organizations developing custom Large Language Models (LLMs) as essential infrastructure. "Recognizing that 400 million people search in Arabic, developers in the Middle East built a regional LLM called ALLaM," he explains. "India did the same with BharatGPT, and other regions are following suit."

• Model maintenance: Owning the entire stack gives an entity the ability to monitor the model for misuse and correct it from the back end," Siddiqui says. "This allows you to intervene proactively before a small problem escalates into a crisis."

But the principle of autonomy extends beyond software, according to Siddiqui. "A dependent supply chain puts your entire strategy at risk."

• Drowning in data: Instead, Siddiqui believes an organization's AI infrastructure must be able to handle the full spectrum of operational demands—including immense data loads from mission-critical functions that dwarf commercial forecasts. "Consider the scale: a large enterprise can generate over 22 terabytes of data per day," he explains. "A single advanced sensor array adds another 100 terabytes daily."

• A quantum leap: However, current technology cannot process this influx of intelligence and sensor data. "We are already receiving data in the hundreds of terabytes daily—soon to be petabytes," Siddiqui continues. "Only quantum machines will be able to perform computations at that speed and scale."

Meanwhile, the immense cost of building quantum-enabled infrastructure will reshape the global landscape and encourage new strategic partnerships, Siddiqui predicts. "We will see new consortia and 'data enclaves' emerge, where enterprises rent secure compute environments. An organization without the capital to build its own AI center can lease a private space. It becomes their secure territory, insulated from the provider's direct influence."

• Access is everything: For Siddiqui, this redefines independence for the digital age. "True autonomy isn’t defined only by where your data resides, but by who controls access to it."

Ultimately, outside control of the underlying infrastructure undermines data ownership and creates a persistent dependency, Siddiqui concludes. "Owning and controlling your own infrastructure, while not feasible for everyone, is the primary way to insulate large enterprises and government from that kind of dependency risk." That's why he advocates for enterprises and governments to build their own dedicated AI infrastructure. "When the business interests of a provider and a client diverge, the provider will always protect its own."

*The views and opinions expressed are those of Shahab Siddiqui and do not represent the official policy or position of any organization or government agency.