Hello, dear readers!

Jensen Huang knows Nvidia cannot rely on GPU dominance forever. That’s why the company is moving beyond chips to expand deeper into AI infrastructure, models, and real-world industries. What exactly is Nvidia up to? Find out below. 

Also, researchers have discovered that internal talent markets have a problem: happy matches are not always high-value matches. That is why talent leaders need to move from unrestricted choice to better-informed internal mobility. More on that below.

This issue features the usual bunch of AI Insights and recommendations for the week's podcasts, books, shows, charts, and tweets, followed by a final chuckle.

Let’s get going.

Industry Highlights: Nvidia Wants to Be More Than the King of AI Chips

For the past few years, Nvidia has been the clearest corporate winner of the AI boom. Its graphics processing units, or GPUs, became the hardware backbone of modern artificial intelligence, powering both the training and deployment of large models. That success turned Nvidia from a high-performing chip company into one of the most valuable businesses in the world. 

But the company’s latest moves suggest it no longer wants to be known mainly as a chipmaker. It wants to become something bigger: a foundational AI company with influence across the entire AI stack.

That ambition was on full display at Nvidia’s annual developer conference, where CEO Jensen Huang unveiled not just new chips, but a broad portfolio of AI systems, models, and infrastructure plays. His message? Nvidia believes the next phase of AI will not be won by selling processors alone. It will be won by controlling more of the technology ecosystem around them.

Opportunity and Threat

Nvidia has enormous financial strength, with huge revenues, strong cash flow, and a cash pile large enough to fund aggressive expansion. Huang’s thesis is that spending on AI infrastructure is still in its early innings, and that trillions more could eventually flow into the sector. If that is true, Nvidia has a chance to deepen its role before rivals catch up.

But it is also driven by threat. Nvidia’s dominance has attracted challengers from every direction. Traditional chip competitors such as AMD are producing alternatives. Startups are designing chips specialized for inference, the stage where AI models answer real-world queries rather than being trained. That matters because inference workloads have different requirements from training workloads, and that opens the door for new architectures.

Even more importantly, Nvidia’s biggest customers are becoming rivals. The hyperscalers, including Alphabet, Amazon, Microsoft, and Meta, buy enormous quantities of Nvidia chips to power their data centers. Yet they are also designing their own custom silicon to reduce costs and optimize performance for their own software environments. If those internal chips become good enough, Nvidia’s grip on the market could weaken over time.

Geopolitics adds another layer of pressure. US export controls have sharply limited Nvidia’s ability to sell its most advanced chips into China. That has created space for Chinese suppliers to grow quickly in their domestic market. Even if those alternatives do not match Nvidia at the high end, they may be sufficient for many use cases. In technology markets, “good enough” can be a powerful competitive force.

Nvidia’s Expansion

Nvidia’s answer is expansion on all fronts. It is moving into other types of chips, including inference chips and CPUs. It is building out its networking business, recognizing that as AI systems scale, moving data efficiently between processors becomes just as important as raw computing power. It is investing in AI models for specific industries, such as robotics, self-driving vehicles, and biomedical research. In other words, Nvidia is trying to own more layers of the AI stack rather than remain concentrated in one.

This strategy also strengthens Nvidia’s case for vertical integration. By tightly linking chips, networking, and software models, the company argues it can deliver better overall performance than a mix-and-match setup from different vendors. That allows Nvidia to sell more than components. It can sell complete systems, increasingly framed as “AI factories,” which are purpose-built environments for AI development and deployment.

New Revenue Streams

The company is also broadening its customer base. Instead of relying too heavily on hyperscalers, Nvidia is pushing deeper into industries such as automotive and pharmaceuticals. Partnerships with companies like Mercedes-Benz and Eli Lilly show its desire to work more directly with end users, not just cloud platforms. That could open new revenue streams, but it could also place Nvidia in more direct competition with the same hyperscalers it still depends on.

Another major lever is investment. Nvidia has become one of Silicon Valley’s most active investors, backing companies in AI applications, software, robotics, and infrastructure. These bets are not just financial. They help create future customers, influence the direction of the ecosystem, and secure important pieces of the supply chain.

None of this guarantees continued dominance. AI spending could cool. Rivals could pressure margins. The market could shift toward workloads better suited to other chips. But Nvidia’s direction is unmistakable. It is no longer content to supply the AI boom. It wants to shape it from the ground up.

The Future of Work: Internal Talent Markets: The Hidden Productivity Trap

As a talent aficionado, I’ve watched internal talent markets become one of the hottest trends in HR. The pitch is simple and appealing: let employees browse open projects and assignments like an internal job board. It promises higher engagement and better retention. But a recent HBR study shows the reality is more complicated than talent teams might expect.

Researchers looked at a large organization that ran both traditional manager-led assignments and a preference-based internal market. A gap emerged:

• When managers assigned people based on company priorities, projected productivity was 33% higher than random matches. 

• When employees got to choose based on their own preferences, the productivity gain dropped to just 5% over random matches, yet workers rated those matches 38% more satisfying.

Happier employees should mean better results, right? Not necessarily. The study found that people often pick roles for reasons that make sense to them personally: better location, flexible hours, or the chance to learn new skills. But these preferences may not line up well with what the business currently needs. Many are chasing broadly transferable skills they can use anywhere, which is smart for their own careers but risky for the company investing in them.

This creates a real tension. A few years ago, when talent was scarce and turnover was high, keeping people happy was the priority. Today, with a softer labor market and more pressure on efficiency, giving employees too much unchecked choice can hurt productivity. The researchers point out that employees frequently don’t have enough clear information about how their strengths match specific tasks or what the company’s real priorities are.

Better Information, Smarter Choices

So, should we scrap internal talent markets? No. But we could redesign them as hybrid systems that still give people a voice while protecting business needs. That means providing better, more transparent information so employees can see how well they actually fit a role. Clear details about required skills, project priorities, and how an assignment aligns with company goals help workers make more informed decisions and reduce bad matches.

Targeted Incentives and Coordination

Even with good information, some misalignment will remain. That’s where targeted incentives come in. Companies can use platforms to nudge strong candidates toward critical roles with rewards like extra time off or special recognition (visible only to those who are a strong fit). At the same time, organizations need to prevent top talent from clustering on the same popular teams. Limits on how many high-performers a single manager can pull in can help spread capability more evenly across the business.

Mastercard’s “Unlocked” platform shows one way this can work. Employees volunteer for extra projects across the business, but those projects have to tie back to real strategic goals. The company has logged over a million hours of this internal work — hours that would otherwise have gone to expensive external hires — while keeping participation voluntary and cross-functional.

Talent Acquisition Perspective

From a talent acquisition perspective, the biggest danger of a poorly run internal market is talent leakage. People use these platforms to build skills, then leave for better offers elsewhere. A smarter hybrid approach flips that dynamic. When employees can find meaningful growth opportunities inside the company (guided by good data and light steering), they’re less tempted to look outside. That means fewer urgent external searches, lower hiring costs, and stronger retention of know-how.

The days of pure employee-led choice may be fading as companies focus more on productivity. Recruiters and talent leaders should be the ones pushing for these balanced systems. Done right, internal mobility becomes a powerful tool: it keeps good people engaged longer while making sure the organization actually gets the performance it needs, from the talent it already has.

AI Insights

• Amazon Acquires Human-Like Service Robots: Amazon has entered the consumer humanoid market with the acquisition of Fauna Robotics, makers of a 42-inch-tall “Sprout” robot with arms and legs that can interact with people, walk, grip items, and dance.

• OpenAI to Discontinue Sora: Despite the popularity of the AI video platform, CEO Sam Altman announced this week that the company would wind down all products that use its video models, including video functionality inside ChatGPT. 

• Amazon Zoox Robotaxi Service to Expand: Zoox will be available for an initial group of users in Austin and Miami later in 2026. The robotaxis feature no steering wheel or pedals.

The Supply Aside

📕 Read - New Cold Wars

David E. Sanger’s New Cold Wars is a fast-paced, well-reported account that draws on deep access to U.S. officials, intelligence sources, and tech leaders to explain how post-Cold War hopes for Russia and China collapsed into today’s simultaneous confrontations. The sections on Ukraine’s battlefields, cyber operations, Taiwan’s semiconductor industry, and White House deliberations make it feel like a genuine first-draft history of this new era of rivalry.

What Else I’m Reading

• Employee Surveys and the “Liar Paradox”: Employee surveys echo the liar paradox: honest criticism suggests a healthy workplace, while widespread negativity may prove the exercise pointless. To avoid breeding cynicism, they need to be well-designed, paired with other feedback tools, timely, and followed by visible action.

• Lower Raise Expectations: In a cooling job market, wage growth has slowed sharply, with many job switchers accepting pay cuts or lateral moves as companies move away from broad raises toward targeted increases for top performers only. Workers should temper their hopes for big bumps and consider job changes more for stability than sizable pay increases.

• Plumbing vs University Dreams: Parents increasingly view skilled trades like plumbing as AI-safe careers with good pay and security, yet only 7% actually want their own children to skip university for them, revealing a  gap between what they recommend for others and what they aspire to for their kids. Deep-seated preferences for academic prestige and social status still steer most toward white-collar paths, even as AI threatens those very jobs.

📺 Watch - On All Fronts With Clarissa Ward

If you’ve seen the disheartening quality of the propaganda videos coming from governments all over during the Iran conflict, you might agree that we need real journalism in conflict zones more than ever. This is an old (2020) interview with CNN’s chief international correspondent Clarissa Ward. She discusses her memoir On All Fronts: The Education of a Journalist, sharing candid reflections on the harsh realities, personal toll, and ethical complexities of reporting from war zones.

👂 Listen - Is the oil crisis about to break global supply chains?

In this Prof G Markets episode, Ryan Petersen of Flexport warns that surging oil prices and the prolonged closure of the Strait of Hormuz risk triggering the worst supply chain disruption in a lifetime, with ripple effects on global trade and costs. His takeaway? The impacts will last a lot longer than the war. Also, Gil Luria explains the latest software stock selloff after Anthropic’s Claude Cowork tool release, while Ed Elson examines suspicious trading patterns suggesting possible insider knowledge ahead of Iran war developments.

🧠 Think: The Pricey Piece of Paper

Thinking about an MBA to boost your career? It is getting more expensive by the year. Tuition at top programs is up 11% over the last four years. And this is not just inflation. It is structural.

Business schools are in a talent war. They are losing professors to tech and Wall Street and paying up to keep them. Full professor salaries are now around $219,000, up 17% in just a few years. Then come the add-ons. Career coaching. Mental health support. Global programs. All valuable, but none of it scales.

That is the core issue. Higher education is a high-touch, people-heavy model. Costs rise because the product itself cannot be industrialized. So, here is the real question. If the cost keeps rising, does the return keep pace? For companies, the answer is becoming clear. Stop paying for the signal. Start building the capability in-house.

Charts of the Week

Quote of the Week

Tweet of the Week

The Final Chuckle

Thanks so much for reading. I’d love to know what you think about this issue and how I can make it more useful to you. If you have suggestions or topics you want to see me address, email me at [email protected]!

-- Naseem