The Alleged $5 Billion Nvidia Chip Trap Remains Unproven
Former CIA officer John Kiriakou and former Russian operative Aliia Roza describe how intelligence operations can use researched intimacy, emotional dependence and isolation to influence a target. Kiriakou argues that a reported Chinese effort involving Nvidia CEO Jensen Huang and a honeypot operative may have sought both access to chips and sensitive information, but the discussion leaves key allegations unresolved. Roza draws on her account of covert training to explain the methods—and their human costs.

The chip allegations remain a chain of claims, not a settled account
John Kiriakou described an alleged Chinese intelligence operation involving Nvidia CEO Jensen Huang and a woman identified in the discussion as Alice Huang. In Kiriakou’s account, Alice represented Megaspeed International, which he said bought about $5 billion worth of Nvidia chips and sent them to China. A graphic shown during the discussion said Megaspeed denied allegations that some hardware was illicitly transferred to China to evade US export controls. The guests treated the alleged operation and its consequences as serious, but the source also contains unresolved questions about what happened and who knew what.
Kiriakou framed the possible strategic stakes this way: China had sought access to Nvidia’s most advanced chips, while the United States had not allowed the company to sell them freely. He defined a honeypot as a relationship in which someone uses intimacy with an important, well-connected person to obtain something. He said Jensen may not have known that the relationship involved a quid pro quo. But, Kiriakou argued, someone close to a powerful executive could also seek information about conversations with the president, trade policy, government agencies, or technology being developed for national-security purposes.
The real danger is the pillow talk that was involved between these two.
The timeline graphic shown in the studio placed Alice’s appointment at Megaspeed in 2023, photographs of her and Jensen together in Taiwan in June 2024, and another public sighting in May 2025. The graphic also marked an investigation reportedly underway by October 2025 and a reported CIA warning in early 2026. These dates formed the chronology the guests discussed; they did not, by themselves, establish what the investigation found or what intelligence officials knew.
Kiriakou questioned Alice’s reported claims to have worked as an anchor for Hunan TV and CCTV. He said Hunan TV had issued a statement that it had never heard of her, and that he had seen no video of her anchoring for CCTV. He also described a New York Times account of a Singapore office with two employees who were not engineers and could not explain what the company did, alongside an empty office in Malaysia. He said these details made him suspect Megaspeed was a front company. The company was described in an on-screen graphic as denying front-company and smuggling claims.
For Kiriakou, the company’s sparse presence suggested “bad tradecraft.” A credible business cover, he said, should have an office, a website, staff who can explain the work, and a record that can withstand scrutiny. He suggested the people behind Megaspeed may have assumed that no one would check before the chips were obtained, or that any scrutiny would come too late. That was his interpretation of the company’s reported footprint, not a finding established in the discussion.
Kiriakou said CNN and the New York Times had reported that Jensen and Alice were inseparable and that she could call him and have him show up. He pointed to their public appearances and said that this did not look like a normal relationship between business partners. He also said Megaspeed was reportedly under investigation by October 2025 for helping China gain access to Nvidia chips.
Kiriakou’s explanation for how an order might have proceeded despite questions about Megaspeed was that Jensen himself could have authorized it. If the CEO told staff that the US government had approved a sale, Kiriakou reasoned, employees might not conduct the scrutiny they otherwise would. This was his theory about how the transaction may have passed through the company.
The timeline graphic and Kiriakou’s account placed a reported CIA memo in March 2026, delivered to the president in early April, warning that Jensen might be compromised by a Chinese intelligence officer. Kiriakou said the warning reflected information someone had brought to the CIA. He compared it to cases in which sources inside one intelligence service warned the other side about a mole. Asked how Alice might have communicated with Chinese officials, he said intelligence services would generally avoid ordinary channels such as phone calls and messaging apps, and might instead use purpose-built communications equipment.
The account’s sharpest institutional question concerned what happened after the reported warning. Kiriakou said Marco Rubio, then secretary of state and national security adviser, had briefed the president. Yet, according to Kiriakou, Jensen later travelled with the president to China and was seated with him and Xi Jinping. Kiriakou said Rubio had initially blocked Jensen from the trip, before Trump called and invited him to join. He said he could not understand the president’s apparent reversal and suggested Trump might simply not have cared. Roza raised another possibility—that Trump knew more or was involved—but presented it as speculation.
Kiriakou identified possible consequences that extended beyond the alleged chip sales: the risk that China could reverse-engineer the hardware, the possibility that private conversations could reveal sensitive information, and investigations by Nvidia’s board and US authorities. He said Jensen had to contend with an internal board investigation as well as FBI inquiries involving the CIA, NSA, and Pentagon. Roza questioned why Nvidia’s security operation had not flagged Alice, citing a reported annual security budget of about $10 million and the public nature of the relationship. Kiriakou’s concern was that a senior executive might carry into a private relationship information about political leaders, trade, defense research, or the government’s thinking.
The guests also speculated about Alice’s reported husband, investor Kang Wei, and his possible role in introducing her to Jensen. They wondered whether someone might profit from a fall in Nvidia’s stock. Roza suggested that a wealthy person could buy shares at a lower price; neither she nor Kiriakou established that an investor had orchestrated events or identified a proven beneficiary. Kiriakou said an operation lasting 15 or 16 years would be unusually long by CIA standards, where officers’ career incentives, in his account, favour results within shorter assignments.
A honeypot is built from research, access, and staged intimacy
Aliia Roza described her training as focused largely on psychology and intimacy. She estimated that 70 to 80 percent of it concerned human psychology, especially men’s, with the rest concerning how to influence a target through intimacy. The method she described did not begin with approaching an influential person and relying on attraction. It began with gathering information: what the target did, where he went, whom he spoke to, and what mattered to him.
That research informed what Roza called love bombing: concentrated praise and attention. In a demonstration with Kiriakou, she did not compliment his clothes or appearance. She praised his work and willingness to speak publicly, and drew on details about his life to establish common ground. She had noticed his Greek background and Orthodox faith, so she greeted him in Greek and mentioned her own connection to Orthodox Christianity. Kiriakou said the connection formed within seconds. Roza’s point was that the praise could feel authentic because she genuinely admired the work she was discussing; in an operation, recognition could help make a target receptive.
The next move might be a small request. Roza suggested asking Kiriakou to review a book draft. The favor would take time and effort, but would also make him feel useful and create a reason to continue the relationship. As she explained it, someone who invests in helping another person may become more inclined to help again. She described this as part of building rapport, alongside remembering small details, attending the same events, and finding shared interests.
Roza said these behaviors had become habitual for her. She did not always experience them as a conscious act; training had become part of how she communicated. She also argued that similar skills could be used outside espionage—in business and relationships, or to help someone feel supported. But the same actions can create a false sense of closeness and make influence harder to recognize.
After establishing initial rapport, Roza described “information control” as becoming involved in the target’s life and studying the people and routines around him. An operative might attend the same events and learn which friends or relatives the target trusts. In her account, the purpose is not only to observe, but to understand where the operative can become useful and what kind of relationship or role the target may be missing.
Roza illustrated a more elaborate form of social engineering with a mission she said targeted a criminal leader named Vladimir. Her team learned that he had been devastated when a woman he loved married someone else while he was away in the army. Roza said the operation used actors in different settings over time to stage scenes of romantic domestic life: a couple kissing, the same-looking couple later appearing together, and eventually the couple with a child. The scenes were spaced out so the target could notice and remember them. The aim, as she described it, was to make a particular kind of future feel imaginable.
Only after that groundwork did Roza approach Vladimir. A team member posing as a small-time drug dealer introduced them. She said she already knew his preferences, schedule, jokes, and behavior; her task was to become the woman he wanted to see again. She adapted her persona to him and tried to make it feel as if the desire to pursue the relationship had come from him.
Roza said the preparation took about two months, followed by an eight-month relationship. She described being with Vladimir around the clock, listening to calls, installing bugs in his home, and reporting what she learned to her commander. She said she passed information through dead drops and meetings arranged under the pretext of attending college. According to her account, authorities used the information to arrest members of the gang in different places where drug supply was taking place.
The emotional bond was designed to produce dependence
Steven Bartlett observed that much of the method sounded like what happens naturally when people fall in love. Kiriakou’s answer was that the operative seeks to control the process; Roza added that the aim is to guide it and influence choices. She described emotional dependency as creating the feeling that the operative understands the target uniquely, that the two are almost soulmates, and that the target becomes his best self in the operative’s presence.
Seduction, in Roza’s account, is less about loving someone than about the emotional state a person experiences with them. The operative tries to shape that state and influence decisions through it. Roza distinguished this engineered attachment from love, describing it instead as obsession or a transactional relationship.
The next stage Roza described was isolation: weakening the target’s ties to people whose judgment might disrupt the relationship. The pressure need not be a direct demand to stop speaking to a sister or friend. Kiriakou and Roza described subtler ways to make a target doubt those relationships: suggest that a sister has criticized the target’s business or image, or is unhappy about the relationship. The operative plants the idea without openly declaring an enemy. Kiriakou said the effect is stronger when the target believes the decision to withdraw came from him.
Gaslighting, in Roza’s account, means making a target doubt his own perceptions and judgment through repeated distortions, partial truths, and planted doubts. In a demonstration, she challenged Kiriakou’s description of honeypot agents as prostitutes, then offered a competing moral frame: an agent serves a state or mission, she said, while a prostitute works for cash. Kiriakou acknowledged that she had changed the frame of the question, while defending his original point as a description of a different context. The exchange made the technique concrete: rather than simply deny a claim, the speaker shifts the categories through which it is understood.
Kiriakou said the goal is to dull a person’s sense of self. Repeating something that may be false or only partly true can make someone doubt an opinion or ability they previously held firmly. Once the target doubts himself, Kiriakou said, he becomes easier to manipulate.
Roza also described mirroring: adapting posture, gestures, vocabulary, voice, and even breathing to the person across from her. In her account, this can create a feeling of connection. She said her training had made such behavior automatic. Kiriakou described a similar persistence in his own work: long after leaving the CIA, he still scanned mirrors and his surroundings for signs of surveillance. Both suggested that training can become ordinary habit rather than a consciously performed technique.
Roza offered caveats about reading these behaviors. Mirroring might be natural, she said, rather than manipulative. She proposed changing one’s own gesture and seeing whether the other person follows as a way of testing the connection. She also warned about someone who seems to know a great deal about you on a first meeting, rapidly matches your interests, and makes the encounter feel like a destined connection. These were her suggestions for things to notice, not proof that a person was running an operation.
A word association became a prompt for further questions
Roza’s “milk technique” is a short word-association exercise. Ask someone what comes to mind when they hear “milk,” she said, and their answers may suggest what emotional register or personal history to explore. Kiriakou answered “childhood” and “mother”; Bartlett said “boobs.” Roza interpreted Kiriakou’s answers as indicating nostalgia and a close connection to his mother, then described a nurturing persona as one possible way to meet that need. Bartlett’s answer, she said, suggested a different role: lover or seducer.
When Bartlett offered “cow” as another association, Roza linked it to stability, food, and support. She described using a response to guide further conversation and look for a role that might appeal to the other person. The exchange showed Roza’s interpretation of the exercise; it did not establish that a single word can reliably reveal someone’s history or needs.
The exercise echoes the broader method Roza described: notice small cues, form an impression of what a person may want, and respond in a way that makes the operative seem to provide it. Her account presented the association as a starting point for conversation, not as a validated profile of the person answering.
The operation’s human costs were not incidental
Roza said she entered the service at 18, after her father, whom she described as a high-ranking officer, took her to a department concerned with human and drug trafficking. She had enrolled to study design, but said she felt unable to refuse her father, who framed the work as a family and national duty. At the academy, she said, she did not know she was being selected for a program involving seduction. She recalled five young women being picked from a class of roughly 350 and trained for covert work; the women had different appearances so they could be used in different operations.
She described herself at the time as inexperienced with men and said her early preparation had focused on psychology, physical training, and weapons. The program’s secrecy meant she could not discuss what she learned with family or friends. Roza said she earned about $100 a month, worked long hours, and was motivated not by money or glamour but by patriotism and the belief that she was serving a larger cause. She described a culture in which dying on a mission could be presented as a privilege or heroic sacrifice.
That account complicates the image of the honeypot operative as a glamorous, freely choosing seductress. Roza said she was abused and raped by commanders, and described suicidal thoughts and therapy. She presented those experiences as part of the reason she now teaches people—especially women—to recognize manipulation and avoid becoming victims of it. She framed her public work as a transformation of the same skills: once used offensively in operations, now used to warn and protect.
Her account of the Vladimir operation adds another complication. Roza said a commander betrayed her by revealing her identity to the criminal gang. She believed the commander wanted to remove Vladimir and replace him with a protégé willing to continue drug trafficking. In her telling, Vladimir wanted to stop the trafficking and legalize his position. The motives and events are part of Roza’s account; the discussion did not independently establish them.
Roza said gang members took her to a forest where she believed they intended to kill and bury her. She said they beat her badly. Vladimir intervened and had her brought to him. She recalled that he asked whether the relationship had been genuine. She told him she had been ordered to seduce him, but also claimed she had fallen in love. He arranged for her to be taken to a safe place. Two months later, she said, she learned from news reports that he had been killed by members of his own group.
Asked whether she felt anything when she learned of his death, Roza answered, “Guilt.” She said two members of her team had also been killed, and that in other operations her role had been to open the door to targets before a separate team carried out killings. She said she had not personally killed anyone, but acknowledged that providing intelligence could contribute to a target being killed. She described lasting pain and a sense of obligation to do something good, and connected that feeling to her stated wish to build an orphanage and adopt children.
Kiriakou shared a different account of responsibility. He recalled being asked by Colin Powell who would be in charge of an operation to kill former US president George H. W. Bush; Kiriakou identified the head of Iraqi intelligence and said 47 cruise missiles were fired at the Iraqi Intelligence Service headquarters in Baghdad. A janitor was killed. His boss told him he had not killed the janitor—Powell had—but Kiriakou said he still felt guilt. Both accounts turned on the distance between providing information or access and carrying out the lethal act.
Tradecraft can persist long after someone leaves the service
Kiriakou described deep cover as a fabricated but durable identity, one that can last decades. A non-official-cover officer might pose as a wildlife photographer, museum curator, or employee of a legitimate company rather than work from an embassy. The job, he said, is to live the cover in public while conducting intelligence work outside ordinary hours. A credible cover needs “backstopping”: an office, colleagues, contact information, and a business story that survives basic checks.
He contrasted that with the Megaspeed setup as he understood it: a company with an obscure presence, few employees, and no clear business activity. Kiriakou said the CIA could arrange a real company or office to support a cover identity, with people prepared to answer questions. He said he had known officers who lived under deep cover for 30 years.
The discussion distinguished a long-term cover life from a “sleeper” arrangement. Kiriakou described a Russian agent, Jack Barsky, who lived in Brooklyn under a fabricated identity. Barsky, he said, started a family and eventually chose not to respond when Russia tried to activate him. When a Russian operative threatened him, Barsky went to the FBI instead. Kiriakou also mentioned a woman arrested in Arlington whom he described as a sleeper agent living as an elementary school teacher.
Roza did not confirm that she had been a sleeper agent. She said she had left Russia officially, then continued to work unofficially at a commander’s request while living in Europe. She said she wanted to leave because the criminal world had learned she was an informant and had tried to kill her. The distinction between formal resignation and continued obligation, in her telling, was not clean: an operative might leave on paper yet remain reachable by former commanders.
Kiriakou also described his own CIA work as the recruitment and handling of sources. The work involved consulting headquarters about how to approach a prospective source, and he recalled asking former officers for ideas when he needed to reach someone outside the usual diplomatic circuit. He said personal connection was a practical instrument: sources could come to believe the officer was a friend or brother, while the officer remained focused on the intelligence relationship.
Asked what motivates people to provide information, Kiriakou said money was the reason in about 90 percent of his experience. Other motives included ideology, excitement, revenge after being overlooked for promotion, and—the additional motive Roza raised—the desire for a legacy. Roza’s account of accepting a low salary emphasized how ideology could outweigh material reward. Kiriakou’s description of a source who cried during a handover captured the divide between emotional experience and professional purpose: the source felt close to him, while Kiriakou said he did not even remember the man’s name.
Red flags are clues to examine, not proof of an operation
Roza’s advice for recognizing manipulation centred on combinations of behavior: unusually rapid intimacy, extensive questions about one’s life, many apparent coincidences, and a new person’s readiness to fill an emotional role. Mirroring alone, she said, can be ordinary. A first-date resemblance or shared interest does not establish intent. Her warning was to notice when these signals accumulate, especially when someone’s attention quickly becomes intense and the relationship begins to displace existing connections.
Her examples show why the tactics can be difficult to distinguish from normal social behavior. Remembering a colleague’s event, praising someone’s work, asking for a small favor, or attending a shared conference can all be sincere. The same actions can be used strategically. Roza’s account points readers toward the pace of a relationship and whether it leaves room for independent judgment and trusted outside perspectives.
Kiriakou said honeypot operations still occur around the world, though he said the CIA generally had not used them for a long time. He named Russia, France, Israel, China, and other states as countries that still used the method. He also cited the FBI’s estimate of 10,000 to 15,000 foreign intelligence officers in Washington. Intelligence work need not target only senior officials, he said. Anyone who can provide actionable information may be useful, including a cab driver with relevant family connections or a staff member who overhears conversations and feels overlooked.
The speakers speculated about whether the reported relationship was part of a larger plot, whether the president knew more than he said, or whether a sudden stock decline might benefit someone. Kiriakou said operations lasting 10 or 15 years were not typical of CIA career incentives, because officers needed visible recruitment results to advance.
The tension between suspicion and evidence surfaced in the discussion of public life. Roza said she still feared consequences for her family in Russia and declined to answer whether she had met Vladimir Putin. She said people continued to call her a spy, sometimes arguing that her public profile could itself be cover. Kiriakou said he faced similar accusations, even though his public criticism of the CIA seemed to him inconsistent with being a plant. Once every public statement can be imagined as cover, suspicion becomes difficult to resolve.
Roza said she now shares what she learned to help people identify manipulation in romantic relationships, families, business, and media. She described her goal as making people more independent of manipulation, and said she hoped eventually to build an orphanage. Her account leaves both the harm of the techniques and their possible defensive use in view: the methods can build attachment, dependence, and compliance, while knowledge of those methods may help people recognize when their judgment and relationships are being narrowed.



