An Open Letter to NVIDIA: I Would Like to Be Your Next CEO

An Open Letter to NVIDIA: I Would Like to Be Your Next CEO

Or: How I Learned to Stop Worrying and Love the Zaibatsu

Dear NVIDIA Board of Directors,

Well, this is a bit awkward.

Jensen Huang is out, your stock has been taking a beating, Congress has questions, regulators have questions, and somewhere in Santa Clara, an extremely expensive conference room is filling with people explaining how nobody could possibly have seen any of this coming. Yeah, I called this months ago.

Given all this, I would like to apply for the position of Chief Executive Officer of NVIDIA.

Yes, seriously.

I recognize that replacing the man who transformed a graphics card company into the economic engine of the artificial intelligence revolution is a rather substantial undertaking. Particularly when your latest reported quarterly revenue is $96.2 billion.

Fortunately, I have spent more than twenty years dealing with complicated technical organizations, ambitious executives, questionable business decisions, and systems that function magnificently right up until someone tries to use them.

I believe I can help. Genuinely.

And unlike the usual parade of financiers, consultants, and professional visionaries who will undoubtedly descend upon your boardroom, I have a slightly unfashionable qualification.

I actually build things. And if you let me, I can build the future of Technology.

I've shipped roughly thirty commercial products and major software releases. I've managed core engineering teams, international development operations, million-dollar programs, and products generating tens of millions in monthly revenue. I've worked across Microsoft, major game studios, live services, security programs, platform ecosystems, and my own companies.

I've designed economies, built production organizations, written technical specifications, published entire game systems, and spent altogether too many nights diagnosing why something that worked beautifully in a presentation collapsed upon contact with a customer.

I even originated a Microsoft developer-platform concept that became Modern.ie and identified more than a million dollars in operational waste by asking the apparently radical question of whether our testing efforts corresponded to what customers actually did.

That last qualification may prove surprisingly relevant.

Because NVIDIA doesn't have a technology problem.

NVIDIA has a systems problem.

And systems are rather my specialty.

First, let's discuss the executive situation.

I have no interest in litigating anyone's private life in an employment application. Whatever the investigations ultimately establish, the governance lesson is already obvious.

An organization responsible for critical technology infrastructure cannot allow its security, commercial relationships, and public credibility to become inseparable from the personal judgment of one individual.

I wrote an essay called The Muskian Design about precisely this species of corporate pathology: brilliant diagnosis followed by catastrophic prescription. The recurring belief that identifying a bottleneck entitles a visionary to become the bottleneck.

I've also worked in Microsoft's security-update and privacy programs. I understand that governance isn't a PowerPoint presentation or a statement of corporate values. It's a collection of accountable people, documented processes, escalation paths, and controls that function even when somebody important would prefer they didn't.

My first order of business would be independent security and conflicts-of-interest oversight, stronger third-party due diligence, documented executive accountability, and an actual succession architecture.

The CEO should never be the company's single greatest point of failure.

This is Systems Design 101, admittedly with rather more expensive consequences.

The objective is to build a corporation that will withstand the test of centuries, rather than planing for the next fiscal quarters.

Then there's your little geopolitical adventure.

Export restrictions have already cost NVIDIA billions in inventory charges and effectively shut it out of significant portions of China's data-center computing market.

Meanwhile, allegations involving chip diversion through third parties have created a compliance and national-security nightmare. Investigations remain ongoing, and accusations should not be mistaken for findings.

But here's what I find fascinating: NVIDIA engineers some of the most advanced tracking, simulation, and computational systems on Earth, yet the industry's physical distribution and end-use verification challenges remain formidable.

I would establish a company-wide export-compliance architecture integrating commercial screening, traceability, partner audits, contractual controls, security review, and escalation authority independent of sales.

And I would stop treating access to China as a dependable assumption in financial planning. We need compliant products for markets we can legally serve, scenarios for markets we cannot, and a strategy that survives geopolitical policy changes. China is the beginning, the global economy is going to drastically shift in the next 15 years and NVIDIA needs to build that infrastructure NOW.

I've spent much of my career turning ambiguous stakeholder demands into executable requirements.

The United States and Chinese governments are admittedly difficult stakeholders. But not wholly foreign ones, either.

Now, about the hardware.

Blackwell rack integration problems. Reported overheating and deployment delays. RTX 50-series driver instability. Missing render output units on some boards. Power connector controversies.

For a company selling the machinery that is supposed to operate the future, this is an unfortunate collection of very expensive present-tense difficulties. I already see paths to solutions, but spilling those beans comes with a prices tag.

My early career involved software and hardware compatibility testing, including Windows, Fable III, and Forza Motorsport. Later, I managed core engineering production for Star Citizen's Alpha 2.6 and 3.0 releases.

You know what I learned?

A product isn't ready because engineering says it works. A product is ready when it survives the conditions under which customers will actually use it.

I would institute stronger system-level qualification, supplier quality gates, field telemetry, cross-component validation, and independent release-readiness reviews. Chip performance, rack cooling, networking, drivers, firmware, and customer deployment would be evaluated as one product.

No more confusing a benchmark with a complete customer experience.

And no, I do not intend to personally redesign a GPU.

That is what your exceptionally talented semiconductor engineers are for.

My job would be to make certain they have the resources, priorities, governance, and organizational freedom to do their jobs properly.

It's an astonishingly simple leadership concept, which may explain why it remains so rare.

We also need to discuss the gamers.

Remember them?

Those strange little creatures who helped build NVIDIA into a household name before the company discovered that selling artificial intelligence infrastructure was rather more lucrative than selling graphics cards to people who just wanted to play Cyberpunk.

Premium pricing, scarce availability, driver headaches, hardware concerns, and GeForce NOW playtime restrictions have all complicated the relationship.

I've written extensively about the dangers of attempting to monopolize customers instead of serving them.

At DIGIT Games, I led feature work associated with an approximately tenfold improvement in conversion on a targeted new-user purchase path in Star Trek Fleet Command. It was an impressive result.

It also exposed a downstream lifetime-value problem: optimizing one metric can distort an entire product economy.

That experience taught me to distrust beautiful dashboards when the underlying customer relationship is deteriorating.

Under my leadership, GeForce would have an explicit consumer-trust program: product reliability, clear specifications, transparent pricing, responsive support, dependable drivers, and service economics that reward loyalty rather than exploit it.

A gamer should buy NVIDIA because NVIDIA is the best choice.

Not because we have successfully exhausted every other choice.

Which brings us to CUDA.

Congratulations. You created an extraordinary developer ecosystem.

You also created precisely the sort of dependency that attracts antitrust scrutiny. Let’s address that.

Regulators have raised questions about market concentration, software lock-in, discriminatory access, and NVIDIA's relationships with cloud providers and competing technology businesses.

I wrote in The Muskian Design that civilization-scale infrastructure cannot safely depend upon a handful of companies whose business model is dependence.

I meant it.

Even when the company in question is the one I would like to run.

My strategy would be to preserve CUDA's technological advantages while investing aggressively in interoperability, open developer interfaces, accessible tooling, transparent commercial terms, and partnerships that strengthen the broader computing ecosystem.

I helped originate Modern.ie because developer friction was a commercial opportunity disguised as a technical problem.

The lesson applies here.

Make NVIDIA the platform developers prefer, not the prison they cannot escape.

There is an important distinction between owning an ecosystem and making yourself indispensable to one. Here in, we create a garden, and invite developers to the safety and free fruit, not lock them in and ration the gruel.

Only the latter is a sustainable competitive advantage.

And then there is the AI money machine.

NVIDIA has invested in customers, suppliers, infrastructure providers, and companies that subsequently purchase substantial quantities of NVIDIA-powered computing capacity.

Critics call parts of this circular financing. You call it strategic ecosystem investment.

Both descriptions can be true, which is precisely why a responsible executive should be concerned.

I have spent years building and analyzing complex game economies. I've watched systems produce extraordinary apparent growth while concealing dangerous incentives and long-term weaknesses.

When the supplier helps finance the customer, who purchases the supplier's product, whose sales justify further financing of the customer, we should probably investigate whether we are looking at durable demand or an extremely sophisticated game of economic musical chairs.

I would separate investment performance from product demand reporting, require independent underwriting for major ecosystem financing, and measure actual compute utilization, customer unit economics, returns on infrastructure, and end-user revenue.

I would also stress-test customer concentration, financing exposure, and a slower AI adoption cycle.

An AI boom is not an excuse to abandon arithmetic.

Somewhere beneath the trillion-dollar valuations, somebody still has to make something useful and persuade another human being to pay for it.

Supply chains, electricity, and the inconvenient physical universe.

Your manufacturing and packaging ecosystem remains heavily concentrated in Asia. Your suppliers face capacity constraints. Your customers need enormous amounts of power, cooling, networking, and capital.

And you increasingly compete with hyperscalers who are developing their own accelerators.

We cannot solve those problems with another keynote.

We can build redundancy through long-term supplier development, multi-regional manufacturing and advanced-packaging partnerships, realistic capacity planning, and broader qualification of critical components.

We can make energy efficiency, infrastructure readiness, and grid investment central product requirements rather than problems discovered after the purchase order.

We should measure useful computational work per dollar, per watt, and per deployment—not merely how many additional GPUs can be packed into a building.

The most successful infrastructure company of the next decade will not necessarily be the one that consumes the most electricity. It will be the one that makes intelligence economically practical. There is power, literally, in diversity.
I am the man that can do that.

Finally, we need to address what AI actually becomes.

I've published my thinking on symbolic AI, persistent memory, relational reasoning, and the limitations of treating ever-larger language models as the inevitable road to machine intelligence.

I've also been developing AI-assisted production systems and experimenting with structured narrative intelligence, local models, and human-reviewed generation workflows.

My position has remained consistent: artificial intelligence needs useful architectures, measurable outcomes, context, memory, efficient inference, and human accountability.

NVIDIA is already investing in robotics, simulation, edge devices, autonomous systems, industrial AI, and open models.

Good.

Now make those investments components of a coherent industrial strategy.

And while we're at it, respect the creators whose work feeds this economy. Copyright and voice-rights lawsuits involving AI training should be treated as warnings about data provenance and licensing, not merely litigation expenses.

As an independent publisher who has written, designed, and commercially produced hundreds of pages of original creative work, I have a fairly direct interest in that distinction.

Technological progress should produce value for the people contributing to it.

Not merely send them an invoice for the privilege.

My Vision: The NVIDIA Zaibatsu

Here is where I become ambitious.

I do not want to make NVIDIA a slightly larger graphics card company.

I want to build the most capable, diversified, and resilient computing-industrial enterprise on Earth.

The ultimate technological zaibatsu.

Not simply a collection of subsidiaries acquired because somebody's investment banker needed another yacht, but a deliberately integrated industrial federation.

A computing division delivering leading processors and accelerators.

An infrastructure division bringing together networking, storage, deployment tools, energy partnerships, and AI factories.

An intelligence division building efficient models, developer frameworks, on-device agents, and secure enterprise systems.

A physical-systems division combining robotics, autonomous machines, industrial simulation, and digital twins.

And a consumer and creator division that makes the results of all this research accessible to actual human beings; not just organizations with billion-dollar procurement budgets.

These businesses would reinforce one another through shared research, open standards, interoperable systems, and disciplined capital allocation.

Our ultimate product would no longer be a GPU.

It would be the capability to transform computation into useful economic activity, anywhere from an individual's desktop to an automated factory.

Silicon. Infrastructure. Intelligence. Industry. Experience.

One interconnected industrial strategy.

And because I know what happens when powerful organizations begin confusing themselves with civilization, the entire structure would need independent oversight, meaningful interoperability, transparent investment decisions, and clear boundaries around market power.

Think the industrial backbone of the next century.

Admittedly, the headquarters can still look like something from a cyberpunk dystopia.

I am not opposed to dramatic architecture.

My First Hundred Days as CEO

I would begin with a board-supervised governance and security assessment, an integrated product-quality and customer-trust review, and a transparent examination of demand, financing exposure, supply-chain dependencies, and infrastructure commitments.

I would establish a cross-functional executive operating system with measurable objectives, explicit decision rights, independent risk escalation, and engineering leadership empowered to challenge executive assumptions.

Then I would work with NVIDIA's existing technical leaders to build a five-year industrial roadmap, identifying where vertical integration genuinely improves cost, performance, reliability, and access—and where partnerships are the better answer.

I would protect the company's extraordinary engineering culture rather than replace it with managerial theater.

I would make succession planning a permanent responsibility.

And I would require every major strategic initiative to answer four questions:

What problem are we solving?

For whom?

What measurable outcome will prove it worked?

And what happens if our assumptions are wrong?

Remarkably, those four questions can prevent a great deal of expensive nonsense.

In Closing

I am under no illusion about the unconventional nature of this application.

I have not previously run a company worth trillions of dollars. I have not led a global semiconductor manufacturer. I would need to rely heavily upon specialists whose technical knowledge exceeds my own.

That doesn't frighten me in the slightest.

Leadership is not the art of being the smartest person in every room.

It is the art of building an organization where the smartest people can do extraordinary work without being defeated by bad incentives, institutional stupidity, and the personal ambitions of the people nominally in charge.

I've spent more than two decades doing variations of exactly that.

I've worked inside enormous organizations and built tiny ones. I've shipped products with substantial resources and products with almost none. I've written about broken systems, designed better ones, and learned some particularly expensive lessons from the difference between what looks good on paper and what actually works.

I believe NVIDIA is standing at a crossroads.

One path leads toward increasingly centralized power, fragile financial dependencies, geopolitical exposure, regulatory hostility, and the eventual discovery that no empire is too large to mismanage.

The other leads toward becoming the indispensable technological foundation of a new industrial age.

A company that makes the next generation of intelligence, simulation, engineering, and automation possible, with the internal discipline not to destroy itself in the process.

And only I can lead you to becoming the latter.

So, NVIDIA Board of Directors, consider this my application.

I am available for an interview.

Already got the jacket.

But first, I would like to see your risk register.

Sincerely,

Ashram Kain
Founder, Product & Technology Executive
20+ Years Building Things That Actually Ship
ashramkain.com

P.S. I am aware that proposing to transform NVIDIA into the world's ultimate zaibatsu may alarm antitrust regulators. I assure you that our legal department will have a seat at the table, preferably before we commission the orbital headquarters.