Last Updated on 10/08/2026
- Cloudflare — Automated internet traffic has now surpassed human-generated traffic, and the company positioned at the center of that shift is seeing renewed growth momentum.
- Optical infrastructure emerges as the next bottleneck — Companies across the optics ecosystem delivered record results amid tightening supply conditions and the possibility of new import restrictions.
- SpaceX — The company posted a standout first-quarter performance while the expiration of a long-standing share lockup removed a major overhang for investors.
Much of the AI conversation typically revolves around the model layer—who has the most advanced models and whether those models can be monetized effectively.
This week, however, markets focused on a different question: who profits from the infrastructure surrounding AI?
Several companies reported results from distinct layers of the AI ecosystem: the internet edge, the physical networking layer, and the frontier of computing. Their earnings represented more than strong quarterly numbers—they highlighted structural shifts taking place across the stack.
- Cloudflare underscored a milestone that may permanently reshape internet traffic dynamics.
- Optics providers posted record performances while supply constraints increasingly resemble the memory shortages seen in previous technology cycles.
- SpaceX released its inaugural earnings report just as concerns surrounding restricted share liquidity began to fade.
The underlying message remains consistent: follow the capital flows. This week, investment attention gravitated toward the network edge, optical connectivity, and low-Earth orbit infrastructure rather than the AI models themselves.
1. Cloudflare: The Infrastructure Behind the Agentic Internet
For the first time, automated and AI-driven traffic on the internet has surpassed traffic generated by humans, and Cloudflare believes this new category could expand another 1,000-fold in the years ahead.
The significance of that milestone extends beyond simple traffic statistics. It signals a transition from an internet centered on people browsing, searching, and clicking to one increasingly powered by AI agents that communicate, make decisions, conduct transactions, and perform tasks autonomously. Cloudflare is strategically positioned at the center of this transformation.
The trend is already showing up in the company’s financial results. Revenue growth accelerated to roughly 36% year-over-year, reaching $696 million, well above Wall Street expectations of around 30% and notably stronger than the low-20% growth rates seen in recent quarters. Other key metrics were equally impressive: remaining performance obligations increased 38%, the company added a record 282 large customers, and net revenue retention improved to 120%, up 600 basis points from a year earlier. Management also lifted full-year revenue guidance to approximately $2.87 billion, helping drive the stock about 16% higher.
Perhaps the most important takeaway was the changing composition of network traffic. AI and machine-to-machine activity now represent more than half of all traffic moving through Cloudflare’s infrastructure, and management expects this segment to grow dramatically from current levels. A company originally designed to sit between users and the internet is increasingly becoming the intermediary between AI agents and the internet itself.
A key beneficiary of this shift is Cloudflare Workers, the company’s edge-computing platform that allows developers to build and deploy applications across Cloudflare’s global network. As AI agents generate ever-larger volumes of requests, infrastructure capable of processing workloads close to end users—or other machines—becomes increasingly valuable.
Developer adoption continues to reinforce that thesis. Nearly 2 million developers joined the platform during the quarter, pushing the total beyond 7.4 million. Workers has become Cloudflare’s fastest-growing product suite, while new monetization opportunities are emerging through services such as paid AI-agent traffic and Cloudflare Pay, a payment framework designed to support the high-volume microtransactions expected in an agent-driven economy.
The takeaway: Cloudflare is evolving beyond its roots as a cybersecurity and application-services provider to become a critical infrastructure layer for the emerging agent-driven internet. In effect, the company is positioning itself as a digital toll collector for AI-powered activity—a software counterpart to the hardware infrastructure companies benefiting from the AI boom. More importantly, its latest results provide some of the strongest evidence so far that the rise of the agentic internet is beginning to translate into meaningful revenue growth.
2. Optics Emerges as the Next AI Infrastructure Bottleneck
If memory chips were the standout infrastructure investment theme over the past year, optical networking technology is increasingly looking like the next major opportunity heading toward 2027. This week strengthened that view through both strong earnings results and a potential regulatory catalyst.
Record Demand Meets Tight Supply
The clearest signal came from Applied Optoelectronics, which reported record quarterly revenue of $191.9 million, up 86% year-over-year, while returning to profitability. Data center demand was the primary growth driver, rising 140% to $107.7 million, with 800G optical transceivers more than doubling sequentially and becoming the largest contributor to sales.
Management projected next-quarter revenue of $255–290 million and expects approximately $1.1 billion in revenue for the full year. More notably, executives indicated that customer orders already extend into mid-2027, with future growth constrained not by demand but by manufacturing capacity and component availability.
That distinction is critical. The challenge facing the industry is no longer generating demand—it is supplying enough hardware to meet it.
The broader ecosystem tells a similar story. Arista Networks raised guidance for the third time this year, while Astera Labs delivered another record quarter with revenue growth exceeding 100%. Together, these results suggest that investors are increasingly rewarding the infrastructure that connects AI systems, rather than focusing solely on the processors powering them.
A Regulatory Catalyst
The investment thesis received an additional boost from reports that U.S. regulators are considering restrictions on Chinese-made optical transceivers used in data centers.
These devices play a vital role in AI infrastructure by converting electrical signals into optical signals and vice versa, enabling high-speed communication between servers, networking equipment, and AI clusters. Chinese manufacturers currently hold a significant share of the global market, meaning any restrictions could redirect demand toward Western suppliers.
Markets immediately focused on potential beneficiaries such as Lumentum and Coherent. However, the larger story may not be market-share gains but supply constraints.
Replacing a substantial portion of global optical-transceiver production is not something that can happen quickly. Manufacturing capacity takes years to expand, and many critical components—including high-performance indium-phosphide lasers—still depend heavily on supply chains linked to China.]
Why the Memory Comparison Matters
The current setup closely resembles the memory industry during previous cycles:
- Demand is growing rapidly and remains relatively insensitive to price.
- Supply expansion requires significant time and capital.
- Capacity constraints create pricing power for suppliers.
- Margin expansion often appears only after shortages become apparent.
In that sense, optics may be following the same playbook that previously drove outsized gains in memory-related businesses. Rather than a temporary demand surge, the sector appears to be facing a structural supply bottleneck that could support stronger pricing and profitability across the value chain for years to come.
The key takeaway is that AI spending is increasingly flowing into the networking and connectivity layers that link computing resources together. As AI clusters grow larger and more complex, optical infrastructure is becoming just as critical as the processors at the center of those systems.
3. SpaceX: A Major Technical Overhang Begins to Clear
SpaceX delivered its first earnings report as a publicly traded company, and the underlying business performance was exceptionally strong.
Revenue surged 92% year-over-year to $7.8 billion, while adjusted EBITDA climbed 191% to $3.5 billion. The company’s net loss narrowed significantly to $541 million, compared with more than $1 billion a year earlier.
Growth Across Connectivity and Compute
The primary growth engine remains Starlink, which generated $4.3 billion in revenue, up 66% from the prior year. The satellite internet business added a record 1.7 million net subscribers during the quarter, bringing its total user base to roughly 12 million.
However, the fastest-growing segment is AI infrastructure. Revenue from SpaceX’s AI and compute operations more than tripled to $2.6 billion, achieved positive EBITDA, and entered the third quarter with approximately $6.7 billion in contracted capacity commitments from customers including major AI developers.
Management reaffirmed an aggressive growth outlook, targeting a $100 billion annualized revenue run rate by year-end and accelerating its timeline toward a $1 trillion revenue objective, now expected by 2030.
During the earnings call, Elon Musk outlined an ambitious AI roadmap, including a commitment to build future AI infrastructure exclusively on Nvidia technology and plans to expand computing capacity beyond 2 gigawatts by the end of this year, potentially approaching 10 gigawatts by late 2027. The company also unveiled StarMind AI-1, an orbital computing platform designed to bring data-center-scale AI processing into space.
Heavy Investment Raises Familiar Questions
The main concern for investors was capital spending. SpaceX reported approximately $18.4 billion in capital expenditures, largely directed toward AI infrastructure expansion. The spending weighed on sentiment and pushed shares lower immediately after earnings.
Yet this concern mirrors a broader debate playing out across the technology sector. The question is not whether demand exists, but whether companies are investing too aggressively ahead of future returns. From an infrastructure perspective, that remains more of a challenge for buyers than suppliers.
The Bigger Story: The Lockup Expiration
The more important development this week was technical rather than operational.
On August 6, a substantial portion of previously restricted SpaceX shares became eligible for trading as the company’s IPO lockup period began to expire. Approximately 911 million shares entered the market, increasing the potential trading float by more than 40% and effectively more than doubling the freely tradable share count.
Many investors had been waiting for this event before establishing positions, expecting the surge in available shares to pressure the stock.
Instead, the opposite occurred.
Despite the anticipated increase in supply, shares rose roughly 6% on the day the lockup expiration took effect. When a widely anticipated supply event arrives and the stock advances rather than declines, it often signals that the market has already absorbed the concern.
For investors who missed the IPO, the lockup expiration may represent a technical reset rather than a fundamental threat.
Key Investment Takeaways
Three separate earnings reports highlighted a common theme: the biggest beneficiaries of AI may not always be the model developers themselves, but the infrastructure layers surrounding them.
- Cloudflare is emerging as a key platform for the growing agentic internet, monetizing the shift from human-driven traffic to machine-driven activity.
- Optical networking infrastructure remains one of the most compelling opportunities, supported by strong demand, constrained supply, and potential regulatory tailwinds.
- SpaceX continues to expand across launch services, satellite connectivity, and AI compute, while the lockup expiration appears to be reducing a major technical obstacle for the stock.
The market remains focused on whether AI spending has peaked. The evidence from this week’s results suggests the opposite: capital continues to flow into the infrastructure that enables AI—across network edges, optical connectivity, and low-Earth-orbit platforms—creating opportunities well beyond the model layer itself.
Leave a Reply