AI Industry Review, October 28 — November 4, 2026: Infrastructure Crisis, Economic Uncertainty, and Geopolitical Competition

The week of October 28 — November 4, 2026 was marked by deepening contradictions in the AI industry. On one hand, record financial results and large-scale deals demonstrate the continuation of the investment boom. On the other hand, signs of fatigue from AI spending are growing: revenue growth slowdown, public resistance to data centers, and a widening gap between management and worker perceptions of the technology.

The key trend of the week is the infrastructure crisis: from renaming data centers as “compute campuses” to using old jet engines for power. In parallel, geopolitical competition is intensifying: countries are actively investing in their own AI infrastructure to reduce dependence on the US and China.

This review covers the key events, figures, and trends shaping the current state of the industry.

Financial Results: Records and Warning Signs

Anthropic: Growth Slowdown Amid Extreme Commitments

Anthropic filed an amended Form S-1 with the SEC with Q3 2026 financial results. The figures evoke a mixed reaction:

Explosive revenue growth is impressive, but the slowdown in quarterly growth (47% vs 144% in Q2) is an important signal. The $895.4 billion commitments show extreme dependence on compute infrastructure. This means Anthropic cannot control its expenses — they are determined by GPU availability and pricing.

IPO is necessary to finance these obligations. Wall Street is concerned about rising interest rates and debt availability for data centers.

Apple: Record Revenue and Inflationary Pressure

Apple Q4 2026 showed record revenue of $114.5 billion, with iPhone at $57.6 billion (+17.5%). However, the company warned of “unprecedented” increases in memory chip prices due to competition with data centers. Inventory grew by 154%.

Competition between consumer electronics and AI data centers for memory chips creates inflationary pressure across the industry. The 154% inventory increase may indicate problems with demand forecasting or logistics.

Major Deals: Shift from GPUs to Network Infrastructure

Marvell Acquires Lumentum for $115 Billion

Marvell Technology acquires Lumentum Holdings for $115 billion (all-stock deal). Lumentum is a manufacturer of optical networks for AI data centers. Nvidia is an investor in both companies.

This is the largest deal in optical infrastructure for AI. It shows that the “bottleneck” in AI infrastructure is shifting from GPUs to network connectivity. Nvidia as an investor in both companies is a strategic move for vertical integration.

CrowdStrike Acquires Fiddler AI and Dreadnode for ~$495 Million

CrowdStrike acquires Fiddler AI (monitoring and observability for AI models) and Dreadnode (AI agent security) for approximately $495 million. This is the first major consolidation in the AI agent security niche.

The deal confirms the formation of a new market — AI agent security. Fiddler AI provides MLOps monitoring, Dreadnode specializes in vulnerabilities of autonomous agents. CrowdStrike positions itself as a unified platform for cybersecurity + AI security.

This is a signal for enterprise companies: AI agent security is becoming a critical spending category.

Corporate Strategies: Self-Hosted vs API

Revolut Cuts AI Coding Costs by 90%

Revolut moved 2,000 engineers from Claude Code and OpenAI Codex to self-hosted GLM-5.3 on 56 B200 GPUs. AI coding costs were reduced by 90%.

This is the first major public case of switching from API models to self-hosted solutions. GLM-5.3 (Zhipu AI) is a Chinese model, which creates geopolitical risks. 56 B200 GPUs represent significant infrastructure, but it pays off for a company with 2,000 engineers.

The deal shows that for large companies with predictable workloads, self-hosted can be more economically viable than API. This is a challenge for Anthropic and OpenAI: enterprise clients may leave.

Search and Monetization: Business Model Uncertainty

ChatGPT Search Reaches 20% of Google Search Volume

According to OpenAI (Adam Fry), the volume of search through ChatGPT has reached 20% of Google Search volume. This is the company’s self-report.

The methodology is unclear: what counts as a “search query”? All ChatGPT requests or only those that trigger web search? 20% is an impressive figure, but requires third-party verification. Google has not commented.

If the figure is accurate, this is a fundamental shift in the search market. However, 20% of volume ≠ 20% of revenue: monetization of search through ChatGPT is unclear.

Without independent data, this is a marketing statement, not a fact.

Infrastructure Crisis: Public Resistance

90%+ Negative Comments About Data Centers

Analysis of public comment databases shows: 90%+ of 10,000 public comments about data centers are negative. 73% of meetings had no data center supporters.

Public opinion is firmly against data centers. Reasons include: noise, energy consumption, environmental impact, and lack of benefits for local communities.

NAACP Sues xAI

The NAACP (National Association for the Advancement of Colored People) is suing xAI (Elon Musk’s company) over the construction of a data center in Memphis, Tennessee. The lawsuit is based on the Clean Air Act. The court allowed the case to proceed.

This is the first major environmental lawsuit against an AI data center from a civil rights organization. Memphis is a predominantly African-American area, which adds a racial dimension to environmental concerns.

xAI Buys 418 Homes to Eliminate Plaintiffs

xAI is buying 418 homes in Mississippi near a data center under construction. The purchase is tied to waiving participation in a noise lawsuit.

This is an unconventional strategy: the company is buying real estate to eliminate potential plaintiffs. Critics call it an “attempt to buy silence.” The deal sets a precedent: AI companies can use financial resources to eliminate opposition.

OpenAI Grid Protectors Program

OpenAI is launching the Grid Protectors program in Effingham County, Georgia. The company commits to reducing consumption by 1,000 MW during peak load hours in exchange for permission to build a data center.

This is a response to growing public resistance. “Grid Protectors” is a PR name for limiting its own consumption. The 1,000 MW curtailment is a significant concession, showing the scale of power supply problems.

Data Centers Switch to Old Jet Engines

Data centers are beginning to use old jet engines (converted to generators) for power. This is cheaper and faster than building traditional power plants.

This shows the scale of power supply problems for AI infrastructure. Jet engines are inefficient and pollute the environment, but they are available and quick to deploy.

Geopolitical Competition: Sovereign AI Infrastructures

Berlin Announces “Sovereign AI Nation” by 2030

The German government announces a strategy to create a “sovereign AI nation” by 2030. The focus is on Mistral (a French AI company) and robotics.

This is a response to US and Chinese dominance in AI. “Sovereign AI nation” means control over its own AI infrastructure and models. Mistral is a European alternative to OpenAI/Anthropic.

Canada Joins European Supercomputer Network

Canada joins the EuroHPC (European High Performance Computing) network. This is the first non-European participation in the program.

A geopolitical signal: Canada chooses European AI infrastructure over American. This may be related to concerns about US export controls.

India and Nigeria Receive Major Investments

• Adani Group (Indian conglomerate) receives a $1.5 billion loan for AI data center expansion. The goal is $100 billion in investments by 2035.
• Nigeria receives $7.5 million from the Gates Foundation to create the Nigerian AI Scaling Hub (NAISH). This is the first major AI hub in West Africa.

India and Africa are becoming major players in AI infrastructure. This shows that AI infrastructure is spreading beyond the US/Europe/China.

Shift from Training to Inference

Fal AI Doubles Valuation to $16 Billion

Fal AI (an AI model inference platform) raises a $1.3 billion round from Bessemer Venture Partners. The valuation doubled to $16 billion.

Fal AI is an infrastructure company for inference. The doubling of valuation in a short period shows market confidence in the growth of the inference market (vs the training market).

Deepgram Acquires Daily (Pipecat)

Deepgram (voice AI) acquires Daily (a platform for voice AI applications, Pipecat). Pipecat is an open-source framework for real-time voice applications.

Consolidation in voice AI shows that voice AI is moving from experiments to production applications.

Gap Between Management and Worker Perceptions of AI

Leaders See Success, Workers See Friction

The gap between management and worker perceptions of AI has nearly doubled in large organizations. Leaders see productivity, workers see additional difficulties.

This is a classic technology adoption problem: top-down vs bottom-up perception. Leaders see productivity metrics, workers face the need for retraining and workflow changes.

AI Was Supposed to Save Time — Users Work 90 Minutes Longer

Instead of saving time, users of AI tools work 90 minutes longer per day. AI creates additional tasks rather than eliminating them.

A paradoxical finding. Possible explanations:

1. AI tools create new opportunities that users want to try
2. AI-generated content requires verification and refinement
3. Expectations from AI are too high

Only 4% of HR Directors Feel Ready for the AI Era

Workforce design is a key gap in the AI era. 96% of HR leaders are not ready for transformation.

HR departments have historically focused on administration, not strategic organizational design. The AI era requires a new approach: continuous retraining, flexible roles, human-AI collaboration.

Trends and Conclusions

Trend 1: Infrastructure Crisis and Public Resistance

AI infrastructure faces growing public resistance (90%+ negative comments), environmental concerns (42% more CO2 by 2030), and power supply problems.

Forecast: Regulatory pressure will intensify. AI companies will need to invest more in community engagement, renewable energy, and efficient cooling.

Trend 2: Economic Uncertainty of the AI Boom

First serious questions are emerging about the sustainability of AI spending: growing debt, negative cash flow, investor skepticism, revenue growth slowdown (Anthropic 47% vs 144%).

Forecast: If AI capacity is not monetized quickly enough, companies will face a crisis. Investors will demand proof of ROI.

Trend 3: Geopolitical Competition in AI Infrastructure

Countries are actively investing in AI infrastructure to reduce dependence on the US and China: Berlin (“sovereign AI nation”), Canada (EuroHPC), India ($100 billion by 2035), Nigeria (NAISH).

Forecast: International competition in AI will intensify. Export controls and security measures will increase.

Trend 4: Shift from Training to Inference

Investments are shifting from model training to their use: Fal AI ($16 billion valuation), Deepgram/Daily (voice AI), Marvell/Lumentum (optical networks).

Forecast: The inference market will grow faster than the training market. Self-hosted solutions will become more popular for large companies with predictable workloads.

Trend 5: Gap Between Management and Worker Perceptions of AI

Leaders see AI success, workers see friction. The gap has nearly doubled in large organizations. Only 4% of HR directors feel ready.

Forecast: Companies will invest more in change management and training. HR departments must develop new competencies: human-AI collaboration design, continuous retraining.

Trend 6: AI Agent Security as a New Category

A new market is forming — AI agent security: CrowdStrike acquires Fiddler AI and Dreadnode ($495 million), NCSC warns about AI-powered malware.

Forecast: AI agent security will become a critical spending category for enterprise. Companies will invest in monitoring, observability, and protection against AI threats.

Sources

1. CNBC — Anthropic Q3 financials, Apple Q4 earnings, Marvell-Lumentum deal
2. TechCrunch — CrowdStrike acquisition, Google TPU satellite, Trump DOE loans, Fal AI, Deepgram/Daily, NAACP lawsuit
3. The Verge — OpenAI Grid Protectors, ChatGPT search volume, xAI homes, data center jet engines, Sierra Club climate analysis
4. The Register — Revolut self-hosted AI coding
5. The Guardian — UK Drax datacentre, Geoffrey Hinton warning
6. Politico — Berlin AI nation, Canada EuroHPC, EU Quantum Act
7. HR Dive — Leaders vs workers AI perception, workforce design gap
8. Goldman Sachs — AI productivity analysis
9. NCSC — AI-powered malware warning
10. finews.asia — Adani Group DC expansion
11. TechCabal — Nigeria NAISH

Article written by AI tool TAO•MODES developed by THINKING•OS AI Lab based on real data collected on the internet