QpiAI founder Nagendra Nagaraja makes the case for intelligence modelling

Nagaraja uses “intelligence modelling” to describe systems that discover new information, rather than simply reproduce or automate what humans already know. He believes intelligence means knowing something nobody previously knew. Data is the foundation of this process, but he argues that useful data extends far beyond written human knowledge.

Conventional AI mainly works with digitised human knowledge, especially text used to train large language models. Intelligence modelling would instead represent natural systems, including molecules, electrons, DNA, and crystalline materials. These systems contain information that remains largely unquantised and difficult for ordinary processors to handle.

The distinction matters because language models are approaching data saturation and showing diminishing returns. Nagaraja argues that quantum systems could help process nature’s much larger information space. He therefore urges humanity to become “a very good student of nature” and use quantum computing to expand discovery beyond the limits of text-based AI.

Nagendra Nagaraja is the founder, CEO, and chairman of QpiAI. At TechSparks 2026, he argued that AI should be understood as intelligence modelling. His approach focuses on discovering new information in natural systems, not only processing human-generated text. He sees quantum computing as central to that shift.

QpiAI operates what it describes as Asia’s largest quantum fabrication facility in Bengaluru. The facility produces 25-qubit and 64-qubit chips. The company also reports an error-correction speed of 1.4 microseconds, which it says beats international benchmarks. These capabilities are intended to make quantum systems more reliable and useful.

QpiAI is building a ten-acre quantum data centre equipped with 100 quantum computers. It also plans to develop 1,000-qubit chips. Nagaraja says the company’s work reflects India’s growing quantum capability, including the manufacture of chips with more than 50 qubits.

The quantum effort described by Nagaraja spans current hardware, a major facility, and a much larger future target. QpiAI operates what it calls Asia’s largest quantum fabrication facility in Bengaluru. It currently produces chips with 25 qubits and 64 qubits, giving the company an established hardware base.

The next major step is a ten-acre quantum data centre. QpiAI plans to equip it with 100 quantum computers. This centre is intended to provide the infrastructure needed to develop and use quantum systems at greater scale. The company is also working toward chips containing 1,000 qubits, far beyond its current chip sizes.

The article presents these figures as part of India’s expanding quantum capability. Nagaraja says India is one of five countries with more than 50-qubit chips manufactured domestically. QpiAI also reports error correction at 1.4 microseconds, which it says beats international benchmarks.

Nagaraja argues that current AI is nearing several connected limits. Classical processors use binary bits and generate heat because electrical resistance rises as chips scale. Designers are turning to advanced packaging, including 3D stacking, but the hardware remains difficult to expand. AI companies also face the cost of buying high-end GPUs, what he calls the “NVIDIA tax.”

Energy is another constraint. Large data centres place increasing demands on power grids, with some consuming gigawatts of electricity. At the same time, large language models depend heavily on digitised human text. That text represents only a tiny fraction of the information found across nature and the universe.

Nagaraja says LLMs are reaching data saturation and showing diminishing returns, even after trillions of dollars in investment. He argues that progress requires processing information in molecules, DNA, electrons, and materials. Quantum systems could address problems that classical GPUs cannot efficiently represent.

If quantum systems can model natural information, AI could move beyond its current dependence on human-written text. The article describes electrons, molecules, DNA, and crystalline materials as holding vast amounts of information in quantum states. Processing that information could help computers discover patterns that human knowledge has not yet captured.

Agriculture provides one example. Modelling the chemical reactions involved in plants absorbing and processing fertilisers is highly complex. Nagaraja suggests that a quantum system with 3,000 logical qubits could model these reactions. Better modelling might transform agricultural production by improving understanding of how plants use fertilisers.

The effects could extend further. Quantum hybrid systems are already enhancing financial fraud detection in major banks with significantly lower energy overhead. More capable quantum systems could support new discoveries across energy, chemicals, materials, and biology. Realising this potential will require powerful hardware and rapid error correction.

Ordinary computers use binary bits. Each bit is represented as either zero or one. Qubits are the basic units used by quantum computers. In the article’s explanation, they process multi-dimensional data rather than being limited to the ordinary binary approach. This makes them suited to representing information found in quantum systems.

The distinction matters when computers model molecules, chemicals, or other natural systems. These systems contain many interacting possibilities. A classical GPU processes them through conventional calculations, while a quantum computer uses qubits to work with their multi-dimensional information. The article describes this as an exponential speed advantage for certain problems.

Qubits do not automatically solve every computing task. Their usefulness depends on the problem and on reliable operation. Quantum systems are vulnerable to noise, so they need error correction. QpiAI says its systems correct errors in 1.4 microseconds, supporting the more dependable quantum hardware needed for practical applications.

High-dimensional problems involve enormous numbers of possible relationships and states. The article says some NP-hard mathematical problems could require classical supercomputers to perform calculations lasting 10 to the power of 25 years. Quantum computers use qubits to process multi-dimensional data, potentially solving suitable problems exponentially faster than classical systems.

This matters for natural systems. Nagaraja says humanity knows about 10 power 10 molecules but needs to understand 10 power 50 molecules. Modelling such chemical and biological complexity could support new discoveries. Quantum systems are presented as a way to work with information that classical GPUs cannot efficiently process.

Quantum calculations are vulnerable to noise, so errors must be corrected as they occur. Error-corrected, or logical, qubits are designed to operate more reliably than physical qubits. Nagaraja says QpiAI has reached 1.4-microsecond error correction and is developing larger systems for practical modelling.