On Monday, July 27, the global semiconductor industry experienced a shockwave that wiped out more than $1 trillion in combined market value within 48 hours.
This panic was not detonated by an economic bubble burst or a kinetic conflict.
The trigger was a single piece of news out of Shanghai about a machine built by a company most people have never heard of, doing a job most people do not understand.
Nvidia lost roughly $250 billion in market capitalization. SK Hynix, one of the biggest memory chip makers fell nearly 14.6% in Seoul, dragging Samsung Electronics down 13.3% alongside it.
AMD, Micron and SanDisk were caught in the blast radius. ASML, the Dutch giant that controls one of the most important technologies on the planet, lost nearly 6%.
This was not the first shock. Months earlier, China's open-source AI models had rattled Silicon Valley. July 27 was the hardware half of the same strategy.
The market simply acknowledged that western export controls, once assumed to be an ironclad firewall, had failed to freeze Chinese progress.
The Monopoly
To understand why markets moved, you first have to understand what markets were protecting.
Today, almost every modern piece of technology, from the smartphone in your pocket to the AI data center requires advanced microchips.
To produce these chips, manufacturing giants like TSMC, Intel, and Samsung rely entirely on Extreme Ultraviolet (EUV) lithography machines built by one company: ASML.
These machines are arguably the most complex commercial products in human history. ASML's newest models, known as High-NA EUV, cost roughly $400 million per unit. It took ASML 17 years to develop this technology, creating an institutional knowledge moat.
ASML's real advantage rests not on one invention, but on an invisible empire of dependencies. The most critical: precision optics.
EUV lithography relies on a 13.5-nanometer wavelength of light to print microscopic circuits. Light at this wavelength is absorbed by almost everything: glass, air, conventional lenses.
So, the industry uses reflective optics coated with up to 100 alternating layers of silicon and molybdenum, each only a few atoms thick.
To prevent the light from scattering, each mirror surface polished to around 50 picometers, smaller than the width of a single atom.
Building these mirrors is extraordinary difficult. Measuring and aligning them is even harder.
These mirrors are produced exclusively by Carl ZEISS SMT in Germany. Zeiss mastered the process through 30 years of relentless refinement.
That is the real moat: not a patent or a clever algorithm, but accumulated engineering know-how so specific and painstaking that it cannot be reverse engineered over a weekend.
Then there is helium. If precision optics represent an engineering problem, helium is a hard supply constraint that no amount of research can overcome.
Semiconductor manufacturing relies heavily on 99.9999% pure helium to carry away extreme heat and purge vacuum chambers.
The United States and Qatar alone produce over 75% of global helium supply. In 2025, China's total helium supply stood at 5,818 tons, comprising over 4,900 tons of imports and less than 900 tons of domestic production.
In March 2026, attacks on Qatar's LNG facilities removed roughly 30% of global helium supply, and Russia simultaneously cut Asian export quotas.
Furthermore, a country cannot build a massive strategic helium reserve to ride out a war like oil. Helium cannot sit in storage indefinitely. It is transported as a cryogenic liquid.
If shipping is delayed, it slowly boils away. This means chip factories can only store a few days' worth of inventory.
That is why the Strait of Hormuz matters to semiconductor markets. Not as a shipping lane but as a chokepoint for the gas that keeps fabs running.
For decades, this is what markets were protecting. On July 27, they started to wonder whether it could still be protected.
Breakout
The answer to the July shockwave lies in a company called Shanghai Aishengna Electronic Technology Group.
Established in August 2023, this secretive state-owned firm absorbed engineering teams from state-backed startups, including Shanghai Micro Electronics Equipment (SMEE) and a Huawei-backed affiliate.
It is not ASML. It is not close to ASML. But it is the first Chinese company to demonstrate it can manufacture DUV lithography equipment at industrial scale.
DUV and EUV are different tools solving different problems. DUV—deep ultraviolet has printed most of the world's chips for two decades.
It is mature and replicable. Any serious semiconductor nation eventually masters it, sanctions or not. What DUV cannot do is resolve transistor patterns below roughly 7 nanometers. That resolution needs EUV. For two decades, ASML was the only entity on Earth that knew how to do it.
On July 27, news broke that Aishengna had successfully begun mass-producing immersion DUV machines using mostly domestic components.
Roughly five machines are planned for 2026, rising to 20 in 2027, destined to major Chinese chipmakers, including SMIC, Hua Hong Semiconductor, and ChangXin Memory Technologies (CXMT).
ASML ships roughly 130 immersion DUV systems per year. This is not a commercial victory against ASML, but it is a proof of concept that China can build the machine without Western permission.
The moment that proof landed, one question became unavoidable. If China can scale DUV under maximum sanctions, what stops EUV?
In late 2025, a state-backed consortium led by Huawei with more than 3,000 researchers across state institutes and private companies built a working EUV prototype in a high-security lab in Shenzhen. The market did not wait for a product launch. It priced in the trajectory: a working EUV prototype, built by the same nation that just demonstrated DUV at scale, is not a science project. It is a timeline.
That left one question the market could not stop asking: how?
The physics bet
ASML generates EUV light using a method called Laser-Produced Plasma (LPP).
It fires 100,000 molten tin droplets per second into a vacuum, then hits each one mid-flight with a high-powered CO2 laser, vaporizing it into a plasma exceeding 200,000°C that emit 13.5nm light which prints the circuits onto silicon.
China cannot source these advanced lasers. So, they chose a different path: Laser-Induced Discharge Plasma (LDP).
LDP uses a ‘Z-pinch’ effect. Massive capacitor banks dump an electrical current through tin vapor, creating a powerful discharge that crushes the vapor with a magnetic field into super-hot plasma emitting the same 13.5nm light.
ASML tried LDP in the early 2000s and abandoned it. Because the violent electrical discharge creates flying tin debris that splatters onto the ultra-precise optical mirrors, degrading them from inside.
These are structural engineering penalties that any Western fab would reject. China is not building the most efficient machine on the market.
It is building a sovereign machine, free from Western control. An inefficient machine producing 100 watts instead of ASML's 600 is an acceptable cost when the alternative is no machine at all.
The talent strategy
China is not guessing at the engineering to close this gap.
The advanced photolithography research group at the Chinese Academy of Sciences', the Shanghai Institute of Optics and Fine Mechanics is led by Lin Nan, previously a research scientist at ASML and its competence owner of light source technology for metrology.
Lin returned to China in 2021 as part of the Chinese overseas high-level recruitment drive—Thousand Talents Program. Before joining ASML, he was mentored by Anne L'Huillier, winner of the 2023 Nobel Prize in Physics.
Now holding over 110 patents related to EUV lithography, he and his team published research in March 2025 demonstrating a laser-to-EUV conversion efficiency of 3.42%—nearing two-thirds of ASML's commercial standard of 5.5%. The gap is shrinking.
This is not reverse engineering; it is the capture of institutional knowledge; the one thing export control cannot stop.
Bipolar tech order
China's hardware success did not happen in isolation, but it landed inside a rapidly hardening geopolitical architecture.
In December 2025, Washington launched the Pax Silica Initiative to secure advanced technology supply chains among what it calls “trusted partners” including the European Union, Japan, South Korea, the United Kingdom, and Israel.
China’s answer arrived on July 16, 2026: the World Artificial Intelligence Cooperation Organization (Waico), established in Shanghai with 29 founding members including Russia, Brazil, Indonesia, Malaysia, Pakistan and South Africa.
UN Secretary-General António Guterres attended the launch. That attendance was not accidental. It was a legitimacy signal aimed directly at the Global South.
The architecture is familiar. In the 20th century, the world divided along nuclear deterrence and competing economic ideologies— Marshall Plan or Comecon, Nato standards or Warsaw Pact specifications.
The Non-Aligned Movement tried to escape the binary. It largely failed, because you cannot run an economy on principled abstraction. The 2026 split follows the same logic through different material. The Global South has watched this movie before.
Waico has no operational track record. What it offers developing nations is not a proven alternative but a proposition: cheaper access with, so far, fewer stated conditions. Whether those conditions remain unstated once infrastructure dependencies deepen is a question Belt and Road recipient nations could already answer from experience.
But the irony runs deeper than the July crash. Washington's original export controls were designed to stall Chinese semiconductor progress. Instead, they became its accelerant. By cutting off access to reliable Western suppliers, Washington eliminated all foreign competition inside China's domestic market. Chinese companies had no choice but to fund domestic suppliers. This dynamic “creative insecurity” forced self-reliance instead of preventing it.
The results are measurable. SMIC posted record revenue in 2025 despite sanctions, and Huawei achieved a 90% domestic substitution rate for components. Furthermore, in October 2025, Beijing invoked its own Foreign Direct Product Rule, explicitly targeting foreign-produced items utilizing Chinese rare earth elements. The controls explicitly target advanced semiconductors and AI research. Trade is a weapon in both directions now. Nobody is pretending otherwise.
The full stack
This is not a chip story. It is the creation of a competing technology civilization.
China already democratized AI through open-source models like DeepSeek, Kimi and others that any nation can use for free. Software came first. Now the hardware layer has caught up. The equation is now complete.
The mathematics of the reality is clear. A developing nation no longer needs to pay Western cloud providers, license Western AI models or depend on ASML for cutting-edge chips. It can build indigenous AI infrastructure today on Chinese tech stack. The technology lags ASML by a generation but for consumer applications, public services and agricultural tools, that gap is irrelevant. For critical infrastructure: power grids, financial clearing systems, defense networks—the answer is different. No government should accept ‘good enough’ on faith when the stakes are national survival.
China does not need to match ASML to win this competition. A domestic EUV machine delivering 150 Watts of stable output would not match ASML commercially, but it would be strategically decisive. If China achieves stable EUV production by 2032, the monopoly is over. Pax Silica's leverage entirely built on the assumption that advanced semiconductor manufacturing cannot be replicated outside a handful of Western-aligned nations, will begin to erode.
The front line
On August 1, 2026, Bangladesh announced its decision to join Waico as an observer.
On that same day, a US Special Envoy Sergio Gor met the Prime Minister in Dhaka and signaled the other door was still open.
But both doors lead to dependency. That is the choice nobody is saying out loud.
The deeper question was never Washington or Beijing. It is whether Bangladesh understands what it is choosing and what it is surrendering before the architecture is locked and the choices become irreversible.
We are at the edge of the most compressed period of change in recorded human history. AI over the next 25 years will unleash more fundamental change than the last two millennia of human history combined.
AI will not change how we communicate or shop. It will change how we think, govern, fight, and raise our children. Every system being built now in healthcare, education, finance, security will be fundamentally reshaped before this decade ends.
This is not a reason for panic. It is a reason for extraordinary caution and extraordinary seriousness.
Bangladesh earned its sovereignty through costs most nations have never paid.
The question is whether that sovereignty extends to the digital layer being laid underneath every other system or whether that layer, quietly, belongs to someone else.
A government that doesn't understand what it is choosing will not get a second chance to choose differently.
The writer is a contributor and a tech enthusiast