On Tuesday, Nvidia agreed to sell hundreds of thousands of its new Blackwell graphics chips to Humain, a Saudi start-up created under the kingdom’s public fund.
The first shipment of 18 000 chips will anchor data centres that the company plans to build with power capacity of up to 500 megawatts. The order dwarfs past shipments to buyers and signals a new scale of demand fuelled by language model training.
AMD signed a $10 billion agreement with Humain to supply central and graphics processors and the software that steers them, according to AMD vice-president Keith Strier. He said the plan keeps Humain from being locked into a single vendor and adds hardware choice for young software firms that will rent the compute service.
Lisa Su, AMD’s chief executive, said the partners are building “a globally important AI platform”. For new AI companies this means easier access to top-tier chips without buying their own racks, lowering entry barriers and bringing Saudi Arabia closer to its goal of acting as an AI hub outside the United States.
Will Lighter US Export Rules Unlock More GPUs For Startups?
On the same day, the US Commerce Department scrapped a Biden-era export rule that would have capped how many advanced GPUs could leave the country without extra checks, according to the department’s guidance. The rule had applied to more than 100 nations, many in the Gulf region that President Donald Trump is touring this week. Startups feared the cap would slow progress on model training at home and abroad.
Chip makers Nvidia and AMD had warned that tight limits might drive clients to Chinese suppliers instead of American ones. Microsoft president Brad Smith told senators that the rule signalled to 120 nations that Washington could not be trusted as a steady source of AI hardware.
European officials applauded the change. Thomas Regnier from the European Commission said member states “represent an economic opportunity for the US, not a security risk” and should buy high-end AI chips without extra paperwork.
For founders in countries that qualify this adjustment turns chip buying from a diplomatic puzzle into a straight commercial deal. Shorter lead times cut costs, and predictable supply lets new labs test models sooner. Paired with the large Saudi data centres, the easier export route could speed up product launches far from Silicon Valley.
How Will The Changes Impact Startups?
Our Experts:
Jes Johnson, Tech Expert, SmartMove
Anjan Pathak, CTO and Co-Founder, Vantage Fit
Elizabeth Lawrence, Managing Editor, MAGA.com
James Dawes, Creative Specialist, Cornish Hue Ltd
Bob Bilbruck, CEO, Captjur
Habibur Rahman, Co-Founder and COO, AIBuster
Jes Johnson, Tech Expert, SmartMove
“Removing restrictions on AI chip exports may sound pro-business, but for startups, it’s a double-edged sword.
“While loosening rules can open up access to global markets and ease procurement bottlenecks, the reality for AI startups is more complex. These changes create short-term flexibility but inject long-term uncertainty into an already volatile space.
“On one hand, startups benefit from easier access to high-performance chips, especially if they’re sourcing from abroad or operating in hardware-limited regions. This could reduce costs and accelerate time-to-market, resulting in huge wins for early-stage ventures.
“However, without stable policy frameworks, many founders may hesitate to scale aggressively or secure VC backing. If export regulations flip-flop with administrations, it’s hard to build a reliable roadmap. Furthermore, larger competitors with established supply chains will adapt more quickly, leaving smaller players exposed to disruption.
“Some may argue that deregulation gives startups more freedom. While that’s true, freedom without consistency leads to fragile growth.
“My advice: diversify suppliers early, monitor policy shifts closely, and factor geopolitical risk into every scaling plan. Startups that stay agile and policy-aware will thrive, regardless of who’s in office.
“In AI, innovation is king, but stability is the crown.”
More from Artificial Intelligence
Amazon Announces Its Own AI Chips To Tackle Growing Demand
Agentic Systems: Software That Thinks on Its Feet
The Evolution Of AI Coding: Why AI Code Review Is The Next Wave
AI Adoption Surges While Governance Lags: Report Warns of Growing Shadow Identity Risk
Can AI And Tech Combat The Loneliness Issue In The UK?
ChatGPT Turns 3: Praise, Problems And A Powerful New Tool
AI Drives $80B Across European Economies, With Sweden Leading The Way
Tovie AI Launches Agent Platform To Bring Scalable AI Automation To The Enterprise
Anjan Pathak, CTO and Co-Founder, Vantage FIt
“From a CTO’s standpoint, lifting the AI chip export restrictions is a significant shift that directly benefits the startup ecosystem. Advanced AI development relies heavily on access to high-performance computing infrastructure, specifically GPUs capable of handling complex model training. Startups, unlike large enterprises, don’t have the luxury of custom-built hardware or exclusive supply chains. These restrictions would have created a bottleneck, slowing down experimentation and innovation at the very early stages where speed is everything.
“By reversing this rule, the U.S. is enabling a more open and competitive global AI environment. Startups can now access the hardware they need to iterate faster, test more aggressively, and compete on a global scale. This move not only removes friction for scaling but also sends a strong message: innovation should not be stifled by restrictive policies, especially when it comes to frontier technologies like AI.”
Elizabeth Lawrence, Managing Editor, MAGA.com
“Trump’s recent moves regarding AI chips will likely have a mixed impact on startups. On one hand, these changes could spark new opportunities by opening up access to cutting-edge technology. Startups in industries like artificial intelligence, cybersecurity, and hardware innovation will likely benefit from lower barriers to entry if the chips become more widely available.
“Though a built-in competitive advantage exists, many startups will still lack the capital and skills to take an active role in these developments. The firms that will be most successful in adopting these new advancements are those with considerable backing and an already established route to distribution. Smaller startups that do not have this will face challenges in competing as larger players will crowd them out for market entry with a far superior capacity to deploy new infrastructure.
“From my reporting, I see clearly that many startups find it difficult to access these types of high-end technologies at either a feasible price or in the necessary business quantities. For example, I have found many startups simply excluded from the consideration of using top-of-the-line hardware, even following Trump’s policies of simplification. Not that there aren’t always folks who manage to extract opportunity from circumstance, but that will not always be possible to realise.”
James Dawes, Creative Specialist, Cornish Hue Ltd
“Well then, from the tech south west in Cornwall, UK – Trump is a western movie echoing across the pond somewhat. However, with his method of board room negotiations and tactics sprinkled with perplexity, to me it’s not all madness. You see, as someone who believes in the free market – it’s important to set tone, risk and momentum under decisions required.
“Take actions and bring people to the table – that’s the art. As the AI boom continues – slight deflationary short-term behaviour isn’t going to hurt that much. More important, it is the nature and outcome that AI can give to humanity – along with a cautious and controlled security blanket.
“As access to AI chips are clearly linked to export systems – countries will now likely face fewer bureaucratic hurdles and regulatory delays when sourcing advanced AI chips. This could accelerate the pace of innovation and reduce operational uncertainty for early-stage companies that rely on cutting-edge hardware. And, less regulation could unlock more growth within innovation, too – really unlocking a more favourable environment for start-ups.
“In the UK, we are a mixed bunch like any country – but, we tend to be later to pop up on the curve of adoption to new waves, especially when issues of environmental or security come to the fore front. It’s the unknown which scares people most I feel – and, everyone needs or wants control. So – like with all technology or political conversations – the average Tom or Peter will still lock themselves out of their mobile phone, scream at the shop staff because of it – but from time to time, will love what technology can do.
“The consensus amidst industry observers is that rescinding the previous rules is “tepid for now” as it removes near-term uncertainty and regulatory burdens. However, the ultimate impact will depend on the specifics of the Trump administration’s replacement policy, which could range from more open trade to targeted bilateral agreements. And, what will the next big tantrum or decision he throws be next? No-one knows – and, that’s half the game.”
Bob Bilbruck, CEO, Captjur
“With Trump removing this AI diffuser rule it just opens up the possibility of them shipping us made AI chips to more potential trade partners outside the US. It really won;t make much of a difference because most of these chips are made in Taiwan now and never ship from the US anyways. NVIDIA, AMD don;t produce these types of chips in the US. These GPU based chips make up most of the AI chip applications today.
“The more relevant story is AI built using CPU based technologies, companies like Symbolic Mind www.symbolicmind.ai who is really going to change the game and CPU chips are produced int he US – so that story is much bigger and much more applicable. It’s important to ask the right questions and have the understanding of how this market will develop, there is not one way to Rome. Many roads lead to ROME in AI developments currently.”
Habibur Rahman, Co-Founder and COO, AIBuster
“Trump’s latest moves on AI chips are of great concern to startups. The export controls target the high-end chips that power sophisticated AI models. Smaller companies will bear the brunt without the chips.
“Speed and compute are the universe for AI startups. They need huge resources to train large models. If they cannot afford the best hardware, they’ll have to settle on their vision or shell out an arm and a leg by renting cloud platforms.
“The impact will not be the same for all startups, though. Some will move to less compute-intensive solutions. Others might turn towards partnerships with bigger companies who still have access to high-end hardware.
“On the flip side, there is potential here as well. Restrictions are often creative. Maybe we’ll be witnessing more attempts at algorithm optimization or building low-resource models that are effective. Sometimes limitations make you smarter and you have breakthroughs that we did not anticipate.
“Adaptability separates winners from losers. Startups that innovate within these limitations will survive and perhaps even prosper. But don’t be mistaken, this is not easy. It’s a rough road ahead.”