Fear is haunting the market. Hormuz escalations are once again translating directly into sticky inflation, pushing the Fed into a tighter corner. During this period, the Semiconductor ETF (SOXX) dropped by almost 30%, with major names like ASML and Micron erasing $66bn and $350bn in market value, respectively. Reflecting on the past three weeks, there are plenty of reasons to be bearish, but what is all this fear actually revolving around?
Breaking the ASML Monopoly
ASML builds one of the most sophisticated machines in the world: the extreme ultraviolet (EUV) lithography machine, widely considered one of the most complex engines ever built in human history. The Dutch company spent over €10 billion and roughly 20 years on R&D just to get the first EUV machine working commercially in 2019. Its extreme engineering complexity, featuring ultra-precise, atomic-level mirrors, creates a technological barrier to entry that is nearly impossible for competitors to overcome, making mere "reverse-engineering" a dream.
Exhibit 1. ASML's EUV lithography machine
Exhibit 2. DUV vs EUV in nanometer precision comparison
On July 8, 2026, China shook the market. A newly named Chinese state-backed entity, Shanghai Aishengna, achieved mass production of domestic immersion deep-ultraviolet (DUV) lithography machines. China reportedly began producing these machines to supply domestic foundries like SMIC, Hua Hong, and ChangXin Memory Technologies (CXMT). Production is expected to start small, with about five units delivered in 2026 and roughly 20 in 2027, pushing Beijing’s effort to be slower amid tightening US-led export controls.
Following the news, ASML's stock tumbled by 4% amid a market overreaction. However, analysts remain skeptical about the rapid pace of China’s lithography ambitions. DUV technology itself functions similarly to the industry standard of 15 years ago. Western sanctions restricted China's access to cutting-edge machines, making DUV immersion their only viable pathway to producing chips.
What are the key components keeping China years behind? It’s not just one part; the reliance on an entire global ecosystem of ultra-precision engineering is what causes the delay. One crucial element is the optics. A German company, Zeiss, produces a near-perfect smoothness that utilizes advanced metrology to measure microscopic flaws. Chinese research institutes like Tongji University and the Shanghai Institute of Optics and Fine Mechanics have reportedly made stunning progress in stitching X-ray optics, but the durability and sub-nanometer precision required for commercial production remain far behind
The 13.5nm light source is also violently complex. ASML’s method involves shooting a high-powered laser at falling droplets of molten tin 50,000 times a second. Replicating this exact tin-plasma laser system is incredibly difficult. Projects led by Tsinghua University aim to use particle accelerators to generate a light source instead. While scientifically viable, building these in massive numbers will take a decade or more to refine into high-yield commercial tools.
Many reports suggest that China already possesses an EUV lithography machine, but it remains a prototype built on a macroscopic scale. Commercially producing the machine requires nanoscale precision. The DUV machines China is currently pushing are only capable of producing everyday chips, such as standard 3D NAND Flash Memory, mainstream DRAM, automotive chips, and consumer electronics. EUV, on the other hand, operates on an entirely different level, enabling AI accelerators, GPUs, advanced data center CPUs, and next-generation DRAM and HBM. China is indeed lagging, but as history shows, they often deliver surprises.
Circular Financing Fear, Yet Again
Previously, we explained how AI capex is reshaping the Nasdaq. The $700bn in investments committed by Hyperscalers will drive their free cash flow into negative territory for the next 10 years, even as semiconductor earnings soar. In recent Q2 2026 earnings releases, names like Alphabet, Amazon, and Microsoft announced even higher 2026 capex. In the first half of 2026 alone, Hyperscalers spent $301bn and are expected to record around $720bn for the year—a 79% YoY capex growth. Even a debt-heavy company like Oracle (ORCL) stated in its earnings release that it is ready to splurge all the cash generated from operations to commit to its capex. 
Exhibit 3. 5Y CDS spreads among tech companies
Exhibit 4. Bond deals issuance 2026
In July, circular financing once again became the dominant fear within the AI bubble narrative. The key to understanding circular financing is revenue round-tripping. Let’s say a cloud provider invests $1 billion in an AI startup, and that startup is contractually obligated to spend that $1 billion on the investor’s cloud servers. The cloud provider then gets to report that $1 billion as "revenue." This artificially inflates the tech giant's financial growth metrics, masking the fact that they are essentially buying their own revenue.
Investors are also flagging default risks now that Hyperscalers are raising cash like there's no tomorrow. In 2026 alone, the number of bond deals exceeding $25bn jumped 3.5 times higher year-over-year. Recently, Nvidia and Oracle's 5Y CDS spreads soared to their highest levels ever. While this AI circular investment loop isn't explicitly stated in the financials, industry watchdogs monitoring multi-year contractual commitments estimate that $1tn in total ecosystem value is bound up in these deals. The primary engine of this loop is the relationship between the major cloud hyperscalers (Microsoft, Amazon, Google) and foundational model startups (OpenAI, Anthropic). A hyperscaler invests billions into a startup; the startup then pledges a significant portion of that capital back to the hyperscaler to purchase specialized cloud computing time.
This situation differs heavily from the 2010s, when chipmakers like Intel, TSMC, and Samsung poured billions into ASML to fund the research and development of the EUV lithography machine. Back then, many analysts worried whether such a massive investment would actually deliver on its promises. The stark difference is that previous investments were betting on the supply chain, whereas current investments are betting on demand. What many investors currently doubt is whether the staggering cost of AI computing is justified; Hyperscalers' massive bets must be validated by ensuring end-user AI demand actually materializes in the near future.
Microsoft Beating Consensus Proves AI Demand
On July 29, Microsoft (MSFT) surprised the market by beating Q4 FY26 EPS consensus by 12% ($4.74 Actual vs. $4.24 Consensus). MSFT booked $90bn on the top line, an 18% YoY increase that crushed Wall Street's expectations. The Intelligent Cloud segment recorded a 31% YoY increase, acting as a true indicator of AI’s immense appetite. This was heavily supported by its Azure cloud computing platform, which skyrocketed by 43% YoY, driven primarily by enterprise customers scaling their AI capabilities. Azure has become crucial for MSFT's earnings, crossing $100bn in annual revenue for the first time and proving that businesses are actively paying for the heavy compute required.
121 Exhibit 5. Microsoft revenue segments fiscal year 4Q26
Exhibit 6.Microsoft (MSFT) stock price YTD 2026
Beyond foundational infrastructure, Microsoft is also proving that AI can be successfully commercialized at the end-user level. The company reported that Microsoft 365 Copilot, its premium AI assistant, has now surpassed 30 million paid seats. This dual-engine growth dispels lingering market anxieties about Big Tech's massive capital expenditures into AI data centers. By successfully monetizing both the underlying compute power (Azure) and daily productivity software (Copilot), Microsoft is demonstrating a complete, profitable AI ecosystem that justifies its ongoing infrastructure investments, the very investments bears were previously shorting.
One week after the earnings post, MSFT stock soared by 28% to reach its YTD high. Microsoft pointed out that it closed its fiscal Q4 2026 with $19.6bn in free cash flow, showing that its AI capex is actually starting to bear fruit. Meanwhile, peers like Alphabet and Amazon slipped into negative free cash flow under the weight of surging infrastructure spending. Bulls are now flocking back to the stock, realizing that Microsoft can fund its infrastructure growth without draining its cash reserves, proving that its Hyperscale position can actually deliver on the AI demand narrative.
Written by: X - bigdigjohnny