A Chinese Spy Inside Nvidia?

The episode opens with the New York Times report that U.S. officials believe Alice Huang—also referred to in the report as Alice Wang or Wang Lee, a Chinese citizen and executive at Singapore-based data-center company Megaspeed—is a Chinese intelligence asset. Officials already suspected she was helping China evade restrictions on advanced Nvidia AI chips, but the more alarming conclusion was that she had allegedly been funneling information to Beijing while maintaining a close relationship with Nvidia CEO Jensen Huang. The reported surveillance supposedly showed that Alice Huang had direct access to Jensen Huang, potentially giving Beijing insight into Nvidia, U.S. policy, and President Trump’s thinking about AI-chip sales and China. Trump was briefed about the findings roughly a year earlier but chose not to act, reportedly because he did not want to disrupt Nvidia’s chip sales, the economy, the stock market, or the U.S.-China relationship. The host notes that the issue could matter when Trump visits China the following month. The Times said Jensen Huang had not received a defensive briefing warning him that he might be targeted by a foreign intelligence service, while surveillance of Alice Huang continued. Nvidia spokesman John Rizzo said meetings with customers were a normal part of Jensen Huang’s job, that he speaks with thousands of executives yearly, and that Megaspeed received ordinary business information—not sensitive trade secrets or national-security information. Nvidia also said it had investigated and debunked earlier claims that Megaspeed was smuggling chips to China. The report did not establish that Jensen Huang knowingly shared classified information.

The host then covers a leak of data relating to more than 41,000 Department of Homeland Security employees and contractors, including personnel at Customs and Border Protection, Immigration and Customs Enforcement, Citizenship and Immigration Services, and the Cybersecurity and Infrastructure Security Agency. DDoSecrets published the archive under the name “Ice Pick” and said it came from DHS’s Okta login system. The files were reportedly last updated on August 19, 2026, and include email addresses, phone numbers, office addresses, internal-system references, and information about undercover and surveillance programs. Wired’s analysis found that CBP employees made up more than half the records, ICE contributed about 10,000, roughly 6,200 belonged to USCIS, and nearly 1,000 were tied to CISA, more than half of them contractors. The data covers about one-seventh of DHS’s workforce and reveals permission groups used to control access to systems. Those groups appear to reference certified undercover operations, wiretaps, communications interception, seized-phone analysis, immigration-court records, and surveillance tools including Palantir’s Falcon and “Pen Link.” They also mention OpenAI’s ChatGPT, Google Gemini, Microsoft Copilot, and other AI tools. The leak’s authenticity and source remained contested. An Okta spokesperson said the company had no evidence its systems or data were compromised, while DHS said it was investigating. National-security expert Alexander Leslie of Recorded Future called the dataset an unusually valuable intelligence resource because it could be combined with commercial and previously compromised information to identify vulnerable personnel, create convincing impersonation attempts, or compromise investigations. The host emphasizes that the danger is not merely exposed contact information but the map of how DHS operates.

DoorDash agreed to acquire Instacart for approximately $17 billion, offering $65 per share—26% above Instacart’s October 22 closing price. Both companies’ boards approved the transaction. DoorDash said Instacart’s brand, enterprise technology products, and app would remain in place, with the two businesses operating separately until the deal closes. The companies said the transaction was expected to close late the following year or in 2028. Instacart CEO Chris Rogers said the company had spent 14 years helping retailers compete and grow online, and that combining with DoorDash would let it do more. The strategic logic is that Instacart brings relationships with more than 2,200 retailers and technology that grocers can use for online ordering and advertising, while DoorDash can provide fulfillment and lower-cost delivery. DoorDash gets a faster route into grocery, where orders are larger and may be more profitable than individual restaurant meals. The deal is also a response to intensifying competition. Instacart was valued at $39 billion by private investors during the pandemic grocery boom and went public in 2023, but its market value had fallen to roughly $13–14 billion. DoorDash, valued at about $73 billion, had already expanded into grocery, including delivery from nearly 2,700 Kroger-owned stores, Family Dollar, and other partners. In the third quarter DoorDash processed 993 million orders, up 28% year over year, with nearly $34 billion in customer spending and $4.6 billion in revenue. Instacart’s second quarter included 90.3 million orders worth about $10.4 billion and slightly more than $1 billion in revenue.

Alphabet’s shares fell as much as 12% after the company reported third-quarter free cash flow of negative $9.07 billion, substantially worse than expected. Revenue rose 25% year over year to $127.6 billion, narrowly missing estimates. Google Cloud beat expectations with $30.7 billion in revenue and grew 103% year over year, but the services business—which includes search—generated $96.3 billion, below the $98.5 billion analysts expected. The pressure is coming from both sides of Alphabet’s AI transition. Capital expenditures reached $51.4 billion in the quarter, up from $44.9 billion in the previous quarter, as Google builds infrastructure and develops AI systems. CFO Anat Ashkenazi said capital spending would increase significantly in 2027, while the company’s projected spending for the current year remained stable. Google is also selling its in-house Tensor Processing Units to customers for the first time, helped by demand for AI infrastructure and cloud services. The description put the full-year capital-expenditure forecast at $195 billion–$205 billion. At the same time, search advertising—the legacy cash engine funding much of the buildout—is slowing. Search-ad revenue grew 12% year over year, compared with 17% in the second quarter and 19% in the first. The host frames the central risk as whether Google can use AI-driven cloud growth and new AI monetization to outrun the decline of its traditional search business. Alphabet has a major advantage because search still produces cash, but the cost of replacing or transforming that cash cow is becoming visible in the financial statements.

Google executives said search queries were growing at their fastest year-over-year rate in five years, partly because AI features are changing how people search. CEO Sundar Pichai said AI Mode queries are three times longer on average than traditional searches, and one in five AI Mode searches globally involves voice, images, or documents. Google also said it had surpassed 50 billion visual searches per month. The company’s argument is that people are using search more often and in richer, more complex ways. The problem is that those new queries currently generate less advertising revenue. Chief business officer Philipp Schindler said AI Mode queries were monetized at a lower rate than traditional searches, although Google was expanding monetization. Google is now scaling ads inside AI Overviews and AI Mode internationally, while adding multi-turn ads in the United States. It is betting that generative AI will create new revenue opportunities even as core search-ad growth decelerates. Google said early results were encouraging: click-through rates for sponsored product listings in AI Mode were comparable to those for organic listings, and user research indicated that people found the ads relevant. The company is extending Gemini-powered formats such as conversational discovery ads and sponsored recommendations within AI-generated answers. The host closes by questioning whether Google risks damaging the product that still pays its bills—an “innovator’s dilemma” in which improving search with AI may make the existing cash cow less profitable before the replacement business is ready.