DENVER — "In my view, GPUs, CPUs and quantum computing are all forms of computing, so they will run together in certain application areas, and you will be able to use quantum computing more efficiently."
Intel Chief Executive Lip-Bu Tan delivered that message on the 15th at the Colorado Convention Center in Denver, taking the keynote stage at .conf26, the annual conference hosted by data management platform company Splunk. According to Jeetu Patel, president and chief product officer of Cisco, Splunk's parent company, who led the keynote session, Tan had attended the U.S. Open tennis tournament that closed in New York the previous day and rushed to the Splunk event after Patel asked him to appear. Tan kept the schedule despite a punishing itinerary that took him back and forth across the country, including a keynote the following day at the artificial intelligence infrastructure summit in Santa Clara, California, and commitments in Boston.
Tan raised a wide range of topics in just 30 minutes. He covered an anecdote about asking Nvidia CEO Jensen Huang for help when Intel was in crisis, the debate over pacing AI investment, the CPU shortage, the growing importance of AI inference, investments in AI chip startups, foundry expansion and power shortages.
Late in the conversation, Tan turned to quantum computing. He brought it up while stressing that the industry as a whole benefits only when companies accept change, even at some risk. "Sometimes companies tend to protect what they have rather than attempt disruptive change," he said. "In quantum computing, people also think the change is taking too long, but I have continued to invest in quantum computing."
Tan's comments point to a future in which quantum computers become the third major class of processor chip alongside GPUs and CPUs, with GPUs and CPUs supporting quantum chips. If GPUs opened the era of AI training and CPUs, deployed alongside GPUs in data centers, drove the era of AI inference, the expectation is that in the quantum era GPUs and CPUs will operate in concert around quantum chips.
In conventional computers, transistor switches represent information as a state in which current flows (on) or does not flow (off), and the computer reads this as 0 or 1 — a bit — processing information in binary. Quantum computers, by contrast, exploit superposition in quantum physics to process qubits, or quantum bits, that hold the states 0 and 1 at the same time, allowing parallel processing. Rather than handling calculations in sequence, they perform them simultaneously on a massive scale, maximizing computing speed. On that principle, a quantum computer can solve in seconds complex calculations that would occupy a supercomputer for hundreds of years.
Quantum computing is regarded as a next-generation technology capable of solving problems beyond the reach of current science, such as drug discovery and code-breaking, but it faces several constraints. Industry and academia currently handle only tens or hundreds of superposed qubits, which limits their ability to tackle real-world problems. Analysts say commercialization will require scaling to more than a million qubits.
When Tan raised quantum computing, Patel asked when he expects it to become a reality. "I am focusing now on error corrections and applications," Tan said. "I also think about who is going to pay for it." The remark signaled that quantum computing is not yet at the point of realization but is in a trial-and-error stage before full-scale use. "There will be far more variations you can work with using qubits," he said. "To me, that is very interesting."
Patel then asked about cryogenic infrastructure, considered the biggest obstacle in the field, and whether Tan sees it as another constraint. "I think it will have a certain impact over the next three to five years," Tan said. Qubits are highly sensitive to heat and vibration, so quantum computers require cryogenic cooling to run without outside interference.
Intel Labs is concentrating on commercializing quantum technology still in the research stage. To that end, it developed Tunnel Falls, a small computing device based on silicon spin qubits that operates at higher temperatures, and is conducting research with academic partners. In December 2020, it unveiled Horse Ridge II, a second-generation chip for controlling quantum computers in cryogenic environments.
GPU makers such as Nvidia and AMD and memory companies including Samsung Electronics (005930.KS) and SK hynix (000660.KS) currently hold the initiative in the AI market, but established players are immersing themselves in quantum research to prepare for the post-AI era. IBM has announced plans to establish Anderon, a foundry dedicated to producing pure quantum semiconductors, and in June it unveiled 0.7-nanometer chip technology, or 7 angstroms, applying a process below 1 nanometer — one billionth of a meter — for the first time in the world. That came five years after it became the first to develop a 2-nanometer chip in May 2021. The aim is to lead the quantum computing era by building chips at the angstrom scale, or one ten-billionth of a meter.
Last month, IBM also announced a new modular architecture for building cryogenic systems. Previously, multiple processors were linked inside a cylindrical cryogenic vessel; the modular approach instead mounts a single quantum chip and connects it to adjacent cells. Removing unnecessary wiring and noise improves stability. It is a core technology for realizing IBM's next-generation quantum computer, due in 2029.
Microsoft unveiled Majorana 2, a chip for quantum computers, in June and said it would commercialize a quantum computer by 2029. Peter DeSantis, who oversees artificial intelligence at Amazon, told CNBC that the first commercially useful small-scale quantum computers will appear within five to seven years.
The administration of U.S. President Donald Trump is actively backing major companies in a race with China to seize the lead in quantum computing. On June 22, Trump signed two executive orders on quantum computing at the White House in Washington. They call for building quantum computing research systems across the federal government by 2028 and deploying quantum sensors in place of the Global Positioning System by 2031. The signing was attended by industry leaders including IBM CEO Arvind Krishna and Ruth Porat, president and chief investment officer of Alphabet, Google's parent company.
The United States is treating quantum technology as a national strategic asset, going as far as acquiring equity in quantum computing companies. Quantinuum said on the 8th of this month that it had signed a final agreement with the Commerce Department for $100 million in support. The department said on May 21 that it would provide $2.013 billion to nine quantum computing companies, including IBM. At the time, it said it would give $100 million each to Quantinuum, D-Wave Quantum and Rigetti Computing, and signaled it would take equity stakes as part of the funding.
