China trained a 1.6 trillion-parameter frontier model on domestic chips. The export-control thesis just took its sharpest test yet β and the results are not what Washington expected.
DeepSeek has released V4-Pro, a Mixture-of-Experts model with 1.6 trillion total parameters β placing it squarely in the tier of GPT-4o and Gemini Ultra by scale. The critical detail is not the parameter count. It is the hardware stack: V4-Pro was trained entirely on Huawei Ascend 950 chips, with no Nvidia H100s, H200s, or Blackwell GPUs in the stack.
This is not a research preview. It is a production-grade frontier model from China’s most capable AI lab, built on China’s most capable domestic accelerators. The US export control regime β which has blocked advanced Nvidia chips from Chinese buyers since October 2022 β did not prevent this.
The timing lands against a busy week for AI geopolitics. At G7, ASML’s CEO said Europe is “quite behind” on AI sovereignty. The US government blocked Anthropic’s Fable 5 from foreign nationals on national security grounds. McKinsey pegged Europe’s AI prize at $1.9 trillion by 2030. And memory analysts are pricing a $94.5B HBM supercycle β a forecast that assumes Nvidia continues to dominate the training infrastructure market. DeepSeek V4-Pro is the most direct challenge yet to that assumption.
The Core Question
“If China can train a 1.6 trillion-parameter model on Huawei silicon, what exactly are export controls controlling?”
The Structural Read
The US export controlstrategy operated on a specific theory: restrict access to the best hardware, and you restrict access to frontier capability. Nvidia GPUs were the chokepoint. Cut off the H100, and China’s labs hit a ceiling. That theory has now produced a testable prediction β and V4-Pro is the data point that challenges it.
Huawei’s Ascend 950 is not equivalent to an H100 chip-for-chip. The performance-per-chip gap is real. What DeepSeek appears to have done is compensate with architectural efficiency β the Mixture-of-Experts design activates only a fraction of total parameters for any given inference call, dramatically reducing compute per token. At 1.6 trillion parameters, MoE architecture means the active parameter count during training runs is a fraction of the headline number. Efficiency, not raw compute, closed the gap.
This matters for the broader AI supply chain thesis. The HBM supercycle β the $94.5B memory forecast that has priced SK Hynix and Samsung into stratospheric valuations β rests on the assumption that Nvidia-architecture training clusters will dominate global AI infrastructure through 2028. An alternative stack gaining credibility in China changes that risk profile. It does not collapse the thesis overnight, but it introduces a scenario that most supercycle models do not price.
MoE architecture activates a fraction of params per forward pass. Efficiency-first design compensates for per-chip performance gaps between Ascend 950 and H100. China’s labs are training at GPT-4 tier scale without US export-controlled hardware.
Economics Signal
$94.5B HBM Supercycle Assumes One Stack
Memory analysts’ supercycle forecast assumes Nvidia GPU dominance through 2028. A proven alternative training stack β even if China-exclusive β compresses the monopoly premium and introduces downside scenarios for HBM pricing that consensus models don’t currently price.
Geopolitical Signal
Semiconductor Sovereignty Is Now Bilateral
The US has chip sovereignty via TSMC and Nvidia. China now has a credible domestic stack: SMIC for fabrication, Huawei Ascend for accelerators, DeepSeek for model development. Export controls accelerated this β they didn’t prevent it.
Three Implications
IMPLICATION 1 β Huawei Ascend Is the Alternative Stack
Before V4-Pro, Huawei Ascend was a credible alternative in theory. Now it is a demonstrated training substrate for a frontier-class MoE model. That changes how every non-US country thinks about its AI infrastructure options β particularly in Southeast Asia, the Middle East, and Latin America, where Nvidia access is restricted or expensive.
IMPLICATION 2 β Export Controls Need a New Theory of Change
The current control architecture targets hardware. DeepSeek’s efficiency innovations β MoE architectures, inference-time compute, distillation techniques β are software and research strategies that reduce hardware dependency. Restricting chips while Chinese labs publish the algorithmic techniques that circumvent the restriction is a leaky embargo. Washington will need to rethink the playbook.
IMPLICATION 3 β Europe’s Sovereignty Gap Just Got More Urgent
ASML’s CEO told G7 that Europe is “quite behind” on AI sovereignty. The subtext of V4-Pro is that the window for building an independent AI stack is closing. The US has one. China now has one. Europe’s $1.9T opportunity (McKinsey) depends on whether it can build β or credibly access β a sovereign infrastructure layer. Dependence on either Washington or Beijing’s stack is a strategic liability at this scale.
This dynamic maps directly onto what we’ve analyzed in AI business model structures β the shift from closed, proprietary AI to open, distributed alternatives is accelerating, and the geopolitical layer is now an explicit variable in every AI infrastructure investment decision. It also connects to the competitive moat framework: Nvidia’s moat has always been the CUDA software ecosystem, not just the silicon. If China’s labs can train frontier models on Ascend, the question is whether CUDA dependency follows β or whether DeepSeek’s stack develops its own ecosystem gravity.
The key insight: Export controls were designed to maintain US AI leadership by restricting hardware access. DeepSeek V4-Pro demonstrates that efficiency-first architecture β specifically MoE at scale β can partially decouple capability from the hardware generation gap. The policy lever worked less well than modeled.
Business Engineer Framework
The AI Supercycle β Who Really Controls the Stack
DeepSeek V4-Pro is a case study in how the AI supercycle plays out when a second hardware stack enters the game. The Business Engineer AI Supercycle framework maps the full value chain β from silicon to model to application β and shows which players hold durable leverage when the infrastructure layer bifurcates.
The US bet that restricting Nvidia would slow China’s AI trajectory. DeepSeek V4-Pro β 1.6 trillion parameters, Huawei Ascend 950, no export-controlled hardware β is the clearest evidence yet that the bet has not paid out on its original timeline. China didn’t hit a ceiling; it built around the wall. The more important question now is not whether export controls failed, but what comes next: a stricter regime targeting algorithms and data, or an acceptance that AI capability is already too distributed to contain. Either answer reshapes the AI investment map for the next five years.
Sources: DeepSeek official release; ASML G7 remarks; McKinsey Global Institute β Europe AI Report 2026; analyst HBM supercycle estimates via industry reports. Published June 19, 2026.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
