Jensen Huang’s comments at Computex defend the Samsung bonus structure that delivers $400,000 to chip engineers, but land alongside an $80bn Nvidia buyback announced two weeks ago.

Jensen Huang, the Nvidia chief executive, told reporters on the sidelines of Computex in Taipei on Tuesday that workers should be paid “as much as possible,” framing the principle as the response to a question about Samsung Electronics’ new bonus structure that delivers as much as $400,000 to memory chip engineers.

“I pay my employees as much as I can,” Jensen Huang said, before adding, “but it doesn’t make this right,” in an unusual public hedging from a CEO who almost never qualifies his own positions in real time.

The Samsung context is the trigger for the question. The Korean memory firm’s union and management reached a deal earlier this month, after a near-strike that survived an injunction filing from a smaller non-chip union, that allocates 10.5% of semiconductor operating profit to chip-division bonuses, with payouts of up to 600 million won (about $400,000) per memory-division worker contingent on sustained profit targets through 2035.

The arrangement was described in Reuters analysis as the largest single profit-share commitment in major Korean corporate history. Samsung supplies HBM4 to Nvidia for the Vera Rubin platform, which made the question to Huang structurally relevant rather than merely topical.

The harder context is the cash-return commitment Nvidia announced two weeks before Huang’s remarks. The company’s Q1 fiscal 2027 results, released on 18 May, included an $80bn share repurchase authorisation, a quarterly cash-dividend increase from $0.01 to $0.25 per share (a 2,400% lift), and a stated commitment to return at least 50% of free cash flow to shareholders through 2026 and beyond. The company returned a record $20bn to shareholders in the quarter alone.

Nvidia’s $81.6bn quarterly revenue and 85% year-on-year growth comfortably support the return profile, but the arithmetic relationship matters: the buyback alone is larger than Nvidia’s total annual payroll many times over.

The two positions Huang is now publicly holding, that workers should be paid as much as possible and that Nvidia should return half its free cash flow to shareholders, are not strictly in tension. Both can be true; the company has the cash to do both.

But the public framing matters at a moment when corporate AI productivity gains are increasingly accumulating to shareholders rather than to workforces, and when Morgan Stanley’s European-banking forecast last week doubled the projected AI-driven job-loss figure to 20%.

Huang’s instinctive defence of high worker pay, even with the “it doesn’t make this right” hedge, is the closest a major AI-infrastructure CEO has come to publicly acknowledging the labour-and-capital tension the AI build-out is producing.

The other Huang comment from this week’s Computex appearances is worth noting alongside. Huang told a separate audience that Nvidia engineers should be using AI tokens worth roughly half their annual salary every year to remain productive, framing non-use of AI tools as analogous to designing chips with pencil and paper.

That position is, on its own terms, a defence of generous worker compensation in token-purchasing rather than salary terms: if engineers are issued $100,000-$150,000 in annual token budget on top of base pay, the realised compensation package is materially larger than the published salary figure suggests.

It is also a framing that depends entirely on AI-token costs remaining at current pricing rather than continuing to rise on the trajectory Commonwealth Bank’s Matt Comyn flagged this week.

Nvidia employs roughly 36,000 people globally. Average compensation per employee, on the company’s most recent disclosures, runs to several hundred thousand dollars including stock-based compensation. The AI boom has driven Nvidia’s share price up roughly 1,170% over the past five years, which has made meaningful numbers of Nvidia employees who hold restricted-stock units into multi-millionaires through normal vesting cycles.

His remark therefore lands inside an Nvidia compensation reality that has, by the standards of large public technology companies, already been unusually worker-favourable.

Huang is travelling to Seoul this week to meet Samsung Electronics chairman Lee Jae-yong and other Korean industrial leaders. The bonus-structure question will likely come up again. The public position Huang has now taken is harder to walk back than the standard executive non-answer.

The ride-hailing company is betting that Germany’s automotive heartland, and a less sensor-heavy approach to autonomy, can finally make robotaxis scale in Europe.

Munich is about to become a test of a particular theory: that the cheapest way to put a driverless taxi on a European street is to stop building special cars for it. Uber said on Sunday that it will launch a robotaxi programme in the German city alongside Autobrains, an Israeli autonomy firm, with the vehicles running on Nvidia’s DRIVE Hyperion platform.

The announcement was made at Nvidia’s GTC conference in Taipei, and the deployment is contingent on German regulatory approval.

The choice of city is not incidental. Munich is the home of BMW and a dense cluster of suppliers, and it offers the mix Uber says it wants: tight inner-city streets, fast ring roads, and what the company politely calls “a thoughtful German regulatory framework.”

Germany has had federal rules permitting driverless vehicles in defined operating areas since 2021, which makes it one of the few European markets where a Level 4 service is a paperwork problem rather than a legal impossibility.

What is genuinely different here is the autonomy stack. Most robotaxis on the road today, the ones run by Waymo and its peers, depend on bespoke vehicles bristling with lidar and a single large end-to-end model trained to do everything at once.

Autobrains is selling the opposite. Its “agentic AI” breaks the driving task into specialised agents, each handling a slice of the problem, running on standard automotive sensors and ordinary automotive-grade compute. The pitch is that this is cheaper to build and easier to drop into any carmaker’s vehicle.

That last point is the commercial idea Uber keeps returning to. The three companies describe the programme as “OEM-agnostic,” meaning the software is meant to run across different manufacturers’ cars rather than a single custom fleet.

“Autonomous driving will not scale by relying on a single model to solve every driving scenario,” said Igal Raichelgauz, Autobrains’ chief executive and founder. “It requires systems that can reason, adapt, and make decisions under uncertainty.”

Uber, which sold its own self-driving unit in 2020, has spent the years since assembling exactly this kind of partnership rather than owning the technology.

It is the same template behind its Tokyo pilot with Wayve and Nissan and its tie-up with Pony.ai and Verne, whose vehicles became Europe’s first commercial robotaxi service in Zagreb earlier this year. Sarfraz Maredia, Uber’s global head of autonomous mobility, framed Munich in the same terms: the hard part, he said, “is bringing them into a commercial network where they can reliably serve riders at scale.”

Several things were left unsaid. The companies named no launch date, no fleet size, and no vehicle. They did not say which carmaker, if any, would supply the cars, nor whether early rides would carry safety operators, as they do in Zagreb.

The Munich plan also tracks a target Uber flagged last year, when it first signalled an intention to begin self-driving operations in the city, so the announcement firms up a timeline more than it sets a new one.

Europe has been the continent where robotaxis are announced more often than they are ridden. Munich is now on the list of places where that is supposed to change, pending a regulator’s signature.

Nvidia opened Taipei’s enormous Computex trade show on Sunday with a spark, literally. The chipmaker unveiled a new PC CPU called the RTX Spark, which it dubbed a “superchip,” and named a who’s who list of PC makers that will soon deliver AI PCs powered by it.

The super-fast, 1-petaflop chip is designed to run AI agents like OpenClaw or Hermes Agent securely, according to Nvidia. Such RTX Spark Windows PCs will be available this fall from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with models from Acer and Gigabyte to follow.

In addition to being equipped with secure sandboxes (jointly developed with Microsoft) to run agents securely, the PCs will also have enough CPU, GPU, RAM, and underlying Nvidia CUDA software to run local versions of large language models.

Nvidia said that its RTX technology will deliver faster performance for AI, better image quality, and support for AI features in more than 1,000 games and applications.

The chipmaker is marketing this as an alternative for creators making AI content, as well as providing a significant upgrade to its traditional market of gamers. Nvidia said more than 100 Windows software makers have signed on to support the new chip, including Adobe, Blender, ComfyUI, Riot Games, and Xbox.

But Nvidia founder and CEO Jensen Huang’s vision for these new PCs is far larger. He wants to end the days of launching apps, pointing, clicking, and typing.

“With RTX Spark and Microsoft Windows, you ask — and the PC does the work,” he said in the press release. “Frontier models. Creative workflows. RTX games. All on a laptop.”

Last month, after delivering another record quarter, Huang promised investors he had found a new $200 billion market for Nvidia in selling CPUs for AI, not just GPUs. He made specific mention of the high-end server CPU released earlier this year called Vera — of which Nvidia says it has already sold $20 billion worth.

He also hinted at his bigger ambitions. “We’ll have billions of agents, and those billions of agents will all use tools. And those tools are going to be like PCs, just like us humans using PCs today,” he said on the earnings call in May. “We’re going to need a lot more CPUs.”

Nvidia ARM-based Windows devices have been tried before — and failed. Back in 2013, Microsoft famously had to write off $900 million on its Nvidia ARM-based Surface RT, with partners like Dell also bailing on the product.

But at this point, after delivering record after record of quarterly revenue, it’s hard to bet against Huang as he pursues his PC dreams once again.

And this chip is an entirely different beast. It’s more powerful, not less. Microsoft is positioning its own RTX Spark PC as so mighty that it named it the Surface Laptop Ultra, and is calling it “the most powerful Surface Laptop ever built.”

Still, PC manufacturers have not released a lot of specifics about each of their offerings, including pricing. These systems appear to be full-fledged Windows versions of the DGX Spark mini-computer that Nvidia already sells to developers for about $4,800.

We’ll have to wait and see if these PCs will compete on price with the affordable Mac Mini that has become a popular choice for running OpenClaw. Or perhaps they will sit at the high end of the PC market, like Nvidia’s own agent-running mini computer.

Either way, if Nvidia has cracked the code on bringing AI agents easily, safely, and usefully to the masses, it could — and should — be big.