Here’s what’s on our plate today:

• ⚙️ Nvidia: no “kill switch” chips—U.S. & China still spars over AI hardware.

• 🧹 Cloudera’s CEO: good AI starts (and lives) with squeaky-clean data.

• 📲 Switch phones this weekend? Apple-to-Google (and back) guide drops.

• 💸 Clay nabs a fresh $100M, soaring to a $3.1B valuation.

Let’s dive in. No floaties needed…

The Laboratory

Decoding Nvidia’s “no backdoors, no kill switches, no spyware” stance

Back in 1998, before novelist Dan Brown was grabbing headlines for his controversial novel The Da Vinci Code, the writer published a novel exploring the concepts of cryptography, cybersecurity, and government surveillance. At the time, the idea that government agencies could keep tabs on users through backdoors in computer software systems was inconceivable.

However, what was inconceivable in 1998 has become a topic of debate in 2025. In what seems like the plot of a thriller novel, Nvidia, the company providing the hardware for the artificial intelligence boom, had to publicly clarify that its chips did not have backdoors or kill switches. The chipmaker also had to urge U.S. policymakers to forgo such ideas, saying it would be a ‘gift’ to hackers and hostile actors.

This story, however, is set in the real world, with real-world implications, and is a stark reminder of the escalating tensions between the U.S. and China and their push to dominate the AI race.

The U.S.-China tussle over technological superiority

The U.S. and China have a history of cooperation and tension. However, in the modern context, the rift between the two nations appeared when the U.S. began restricting the export of some high-end microprocessors that were crucial to the manufacturing of powerful AI systems.

The U.S. cited risks to national security, saying closeness of companies like Huawei, which were spearheading China’s tech rise, to its communist government raised risks of espionage and use of advanced technology in military applications.

With AI bursting onto the scene, matters escalated quickly. Between 2024 and 2025, the U.S. pushed for allies like the Netherlands and Japan to limit exports of chipmaking equipment.

The U.S. export curbs were focused on ensuring China did not have access to Nvidia’s A100 and H100 GPUs, which are crucial to powering its AI ambitions. However, while the U.S. was increasing restrictions, Chinese companies were busy finding novel ways of ensuring they had access to the GPUs needed to train and run advanced AI models.

While Chinese startups like DeepSeek claimed they managed to train models with older GPUs, many doubted the claims, suggesting that the company had secretly accessed and used the latest Nvidia H100 GPUs.

The Chinese government also pledged billions to boost its domestic chipmaking capabilities to reduce reliance on imports.

U.S. policy U-turn: Strategic retreat or tactical maneuver?

In July 2025, signaling a significant policy shift, the Trump administration relaxed chip export restrictions to China, aiming to prevent China from accelerating domestic chip production and overtaking the U.S. in semiconductor technology.

According to a Reuters report, the White House National Economic Adviser, Kevin Hassett, said, "President Trump and his team decided to let the NVIDIA chips go" to maintain America's technological edge in AI chips.

Behind the scenes, the relaxation is part of wider negotiations between the two countries for the supply of rare earths, which is largely controlled by China. The supply of rare earths is critical to modern technology and impacts sectors like AI, GenAI, robotics, space exploration, clean energy technology, and rocketry, all of which depend on them. China reportedly controls 68% of the global rare earth mining and processing, and has been leveraging this position to pressure the U.S. to ease off on its curbs on chip exports.

However, despite successfully leveraging its critical role in global supply chains to secure concessions, China remains wary of American hardware.

China’s skepticism over U.S. concessions

Despite having leveraged its position as the main supplier of rare earth globally,  China was not convinced of U.S. motivations.

Soon after the export curbs were lifted, the Cyberspace Administration of China (CAC) summoned Nvidia to explain whether its H20 chips have any “back-door safety” risks. The regulator also requested that Nvidia provide explanations and relevant proof concerning those issues.

And, China’s concerns over the security of Nvidia chips are not unfounded.

The CAC’s inquiry, it said, was based on reports of security issues involving Nvidia chips. The regulator also pointed to demand from U.S. lawmakers to verify the location of artificial-intelligence chips like those made by Nvidia after they are sold.

The Chinese government fears that backdoor functions, including geo-tracking and remote controls, embedded in American hardware could compromise sensitive data and lead to device vulnerabilities and potential economic losses.

Meanwhile, demands for geolocating chips have bipartisan support in the U.S. as lawmakers try to address reports of widespread claims that China is smuggling chips in violation of export curbs.

What are kill switches and backdoors?

The Chinese government suspects Nvidia chips could potentially have covert software features that would allow bypassing normal security controls to gain access to devices. Or that the U.S. could force the company to design a kill switch in the chips. These are built-in mechanisms in the hardware that allow them to be remotely disabled in case of emergency.

However, kill switches can be misused or hacked, rendering critical systems inoperable.

The U.S. lawmakers' push to incorporate geolocation features in AI chips is akin to having spyware that would allow adversaries to track and potentially target critical infrastructure.

Nvidia rejects kill switches and backdoors

While the U.S. and China use high-end GPUs and rare earths as bargaining chips to secure their motives, Nvidia has had to come forward with its stance.

The company, in a blog post, said, “Some pundits and policymakers propose requiring hardware 'kill switches' or built-in controls that can remotely disable GPUs without user knowledge and consent.” This, the chipmaker says, currently does not exist, and should not exist.

Nvidia highlights lessons learned from the 1990s Clipper Chip initiative. That program, led by the NSA, included chips with built-in encryption and government access via a key escrow system. However, it ultimately exposed vulnerabilities exploited by malicious actors, underscoring why embedding backdoors creates long-term risks.

Additionally, Nvidia says that hardwiring a kill switch into a chip would be a permanent flaw and an “open invitation for disaster.” As such, the company is calling for governments to refrain from deliberately weakening critical infrastructure.

Can Nvidia’s assurances restore trust?

Nvidia’s stance underscores the deepening distrust in the global tech supply chains, especially amidst the rising U.S.-China tensions. As the countries compete to win the AI race, Nvidia finds itself in a unique position. The race for AI dominance, while boosting its revenue, is also putting it in the center of this geopolitical chessboard, forced to assert its neutrality and technological integrity.

The chipmaker’s public stance opposing kill switches and backdoors is not just a corporate defense; it is a call to preserve the reliability and security of global AI infrastructure. And as AI adoption increases, ensuring neutrality in the geopolitical landscape will determine if future AI systems will help build a more transparent and collaborative global order, or one that relies on surveillance and distrust to survive.

Quick Bits, No Fluff

• Cloudera CEO, Quality Data Fuels AI Payoffs: Rob Bearden says “flashy models flop when they’re trained on garbage in.”

• Google Woos iPhone Owners Back: New Android tools mimic iOS favorites and promise painless photo-chat migrations.

• Clay Snags $100M, Hits $3.1B Valuation: The ‘relationship-intelligence’ startup will double down on gen-AI inside its CRM killer.

Three Things Worth Trying

• 🧩 Perplexity Pages – Hit ‘Create Page’ and watch an AI co-author polished research briefs—complete with citations—for your next deck.

• 🗂️ Amplemarket AI Sequences – Paste a LinkedIn URL; it drafts a multi-touch outbound cadence that actually sounds human (and syncs to HubSpot / Salesforce).

• 📸 Scenario ‘Instant LoRA’ – Fine-tune Stable Diffusion on <100 images in minutes—great for marketing teams that need brand-consistent art without a 3-day render.

Prompt Of The Day

Meme Of The Day

Thursday Poll

Rate This Edition