Are We in an AI Bubble?

There’s a growing debate about whether we’re in an AI bubble. If we are, how will we know?

The Case For a Bubble

There are growing signals that the AI, and by extension tech-investment, phenomenon may be inflating into a bubble. For example:

• The Bank of England recently flagged a risk of a “sudden correction” tied to the booming AI sector.
• The Richmond Fed notes that AI usage and expectations among businesses have surged (e.g., from ~3.7 % to ~10 % recently), reminiscent of past tech-boom dynamics.
• A recent analysis points out that valuation expansions are already very pronounced, for example, power providers supporting AI infrastructure now trade at significantly higher price-to-sales ratios than a few years ago.
• One commentary argues the “bubble” size may be enormous, driven by heavy debt and capital expenditure in AI infrastructure, raising red flags about sustainability. 

In short: rapid investment, soaring valuations in companies with little track-record of profitability, and comparisons to the dot-com era all feed the bubble narrative.

The Case Against a Bubble

On the other hand, there are arguments that what we’re seeing isn’t a classic bubble, or at least not one that’s about to burst imminently. For example:

• Massive, real revenue growth: OpenAI reportedly generated about $4.3 billion in the first half of 2025, already surpassing its full 2024 revenue of $3.7 billion and on pace for $12–13 billion this year. Anthropic, meanwhile, is targeting $9 billion in annualized revenue by the end of 2025 and up to $26 billion by 2026, reflecting broad enterprise adoption of its Claude models. These numbers suggest genuine market demand, not speculative hype.
• Some analysts suggest that AI is a fundamental, structural shift in technology and productivity, not just speculative hype, which could justify elevated valuations.
• Others note that while some “AI-exposed” companies may be overvalued, the broader ecosystem is tied to real business spending, infrastructure build-out, and long-term demand. For instance, survey data shows many businesses expect to deploy AI in the near term.
• The historical research (e.g., comparing dot-com and AI eras) suggests that the traditional markers of a bubble are not mapping cleanly onto the current AI wave, meaning we might be in something new rather than a repeat of the 2000 dot-com collapse.

Thus:  maybe this is not simply a "bubble" but a complex mix of hype + real transformation; the question is how much is hype and how much is sustainable.

Conclusion

We’re clearly seeing bubble-like dynamics in the AI sector: soaring valuations, debt-fueled infrastructure spending, and lofty claims that often outpace near-term profitability. However, this doesn’t necessarily mean an imminent collapse. The likely scenario is a market segmentation, where weaker players and hype-driven ventures fade while companies with real technological depth and revenue strength endure.  

Still, investors should remain alert to key warning signs: excessive valuations unsupported by earnings, leveraged expansion, inflated expectations, and over-concentration among a few tech giants. If these factors persist alongside weak returns or a sudden shock, such as regulatory changes or shifts in sentiment, the correction could come quickly. For now, AI remains an overheated but transformative sector, one that may cool unevenly as reality catches up to ambition.

And now, here is the AI news from last week.

Not a subscriber yet? Click here to subscribe.

Notable AI news from Last Week

Could AI Bubble 'Crowd Out' Other Parts Of U.S. Economy?  As Amazon, Microsoft, Google, Meta, and Oracle prepare to report third-quarter earnings, investors are looking for clues on 2026 capital spending amid an AI-driven infrastructure boom now nearing $400 billion. Economists are divided on whether this surge represents innovation or an early-stage “AI bubble.” Some, like Ed Yardeni, argue that tech giants are self-funding their AI buildouts and that fears of “crowding out” other industries are exaggerated. Others, including SK Ventures’ Paul Kedrosky, warn of parallels to the 1990s telecom bubble, when excessive capital concentration hurt U.S. manufacturing competitiveness.  

Top OpenAI, Google Brain researchers set off a $300M VC frenzy for their startup Periodic Labs: Periodic Labs, founded by former OpenAI researcher Liam Fedus and ex-Google Brain scientist Ekin Dogus Cubuk, has emerged from stealth with a $300 million seed round led by Andreessen Horowitz and Felicis. The startup aims to merge large language models, robotic labs, and high-fidelity simulations to accelerate the discovery of new materials such as superconductors. By integrating real-world experimentation into the AI training loop, Periodic aims to create a continuous feedback loop in which failed experiments are as valuable as successes, potentially redefining how scientific data fuels model learning. 

Anthropic Has a Plan to Keep Its AI From Building a Nuclear Weapon. Will It Work?  Anthropic has partnered with the U.S. Department of Energy and the National Nuclear Security Administration to ensure that its AI model, Claude, cannot assist in building nuclear weapons. Using Amazon Web Services’ classified cloud, the agencies ran red-team tests on Claude and co-developed a “nuclear classifier,” a filter trained to detect conversations that veer into nuclear-sensitive topics. While Anthropic positions this as a proactive safety step and offers the tool to other companies, critics argue it may amount to security theater since Claude was never trained on classified nuclear data.  

The FTC Is Disappearing Blog Posts About AI Published During Lina Khan’s Tenure:  In late July 2024, FTC chair Lina Khan spoke at Y Combinator, advocating for open-source and “open-weight” AI models that would let smaller developers compete more easily. The speech coincided with debate over California’s proposed SB 1047 safety bill, which Khan opposed as too restrictive. Soon after the Trump administration took office, the FTC quietly deleted Khan-era blog posts that supported open models and warned of AI-related consumer harms.  

AI Is Changing What High School STEM Students Study: Once considered the golden ticket to a secure future, “learn to code” no longer carries the same allure for students entering the AI era. As generative AI begins to automate programming tasks, high school and college students are pivoting from traditional computer science toward statistics, data literacy, and applied mathematics—fields that emphasize interpretation and reasoning over rote coding.

OpenAI’s Atlas Browser Takes Direct Aim at Google Chrome: OpenAI has unveiled Atlas, a new AI-powered web browser that embeds ChatGPT directly into the browsing experience. With a sidebar for asking questions about any webpage, an autonomous agent that can click and complete tasks, and a chatbot-driven search interface that replaces traditional link lists, Atlas reimagines how users interact with the web. The browser, now available globally for macOS and coming soon to Windows and mobile, also includes an optional “browser memory” that lets ChatGPT recall past searches and automate recurring tasks. CEO Sam Altman called Atlas a once-in-a-decade opportunity to reinvent the browser itself, positioning OpenAI directly against Google’s AI-enhanced Chrome and emerging rivals like Perplexity’s Comet.

Recent Investment Activity

Tracking 49 AI companies last week that raised $3.2B, here are the highlights: 

1001 AI — Dubai startup building AI infrastructure to optimize operations in aviation, logistics, construction, and more. Raised a $9M seed co-led by CIV, General Catalyst, and Lux Capital. 

Acelab — NY platform that helps architects/designers research and select building materials with AI. Closed a $13.5M Series A led by Navitas Capital.

Acoru — Madrid startup combating AI-driven fraud and AML for banks. Raised an $11.6M Series A led by 33N Ventures with Adara and Athos.

AdsGency — SF company whose AI agents plan, produce, and manage multichannel ad campaigns. Landed a $12M seed led by XYZ Venture Capital, with Streamlined Ventures, HF0, and Hat-Trick Capital.

Anchor Browser — Tel Aviv infrastructure enabling AI agents to carry out web tasks autonomously (beyond API walls). Secured a $6M seed co-led by Blumberg Capital and Gradient Ventures.

Anrok — SF platform using AI to automate global sales tax compliance for digital businesses. Raised a $55M Series C led by Spark Capital; total financing now $100M+.

Bexorg — New Haven company combining AI with whole-human brain science for CNS drug discovery. Closed a $23M Series A led by Engine Ventures (total raised $42.5M).

Bricklayer AI — Arlington, VA tools to automate security operations centers with AI. Raised a $5M seed led by Tech Square Ventures (total $7.5M).

Bronto — Dublin startup for AI-era log management and analytics. Closed a $13.9M seed led by Cercano, with Heavybit and Conviction Capital.  

Chemify — Glasgow company using AI + robotics to automate molecular design and synthesis for drugs and materials. Raised a $50M+ Series B co-led by Wing VC and Insight Partners.  

ChipAgents — Santa Barbara startup automating chip design and verification with agent-based AI. Secured a $21M Series A led by Bessemer; strategics include Micron, MediaTek, and Ericsson.  

Coolant — Cambridge, MA, spatial-AI tools to monitor and optimize land and natural resources. Raised $4.3M across seed and pre-see,d co-led by General Catalyst and Floodgate.  

Counsel Health — SF company using AI to triage health questions and route cases to physicians. Raised a $25M Series A co-led by Andreessen Horowitz and GV.  

Crusoe — Denver builder of data centers and energy infrastructure for AI workloads. Closed a $1.4B equity round at a $10B+ valuation, co-led by Mubadala Capital and Valor.  

CurbWaste — NY platform digitizing waste-hauler operations with AI for routing, payments, and inventory. Raised a $28M Series B led by Socium Ventures (total $50M).  

Darwin AI — NYC startup helping government agencies manage AI compliance and governance. Raised a $15M Series A led by Insight Partners (total $20M).  

Defakto — Palo Alto identity and access for non-human actors (agents, services) to secure automated interactions. Closed a $30.75M Series B led by XYZ Venture Capital.  

Dialogue AI — Los Angeles market-research platform powered by AI. Raised a $6M seed led by Lightspeed, with Seven Stars, Uncommon Projects, and Tornante.  

ExaCare AI — NY tools to automate admissions and clinical decisions for post-acute care. Closed a $30M Series A led by Insight Partners.  

Expedition Medicines — Cambridge, MA, drug discovery using gen-AI and quantum covalent chemistry for oncology/immunology. Raised a $50M seed led and incubated by Flagship Pioneering.  

Findem — SF AI talent platform to identify and manage job candidates. Raised a $51M Series C led by SLW (total $105M).  

Finster AI — London software to automate research and document workflows for banks and asset managers. Closed a $15M Series A led by FinTech Collective.  

Fourier Health — Miami AI system that summarizes and organizes patient data for clinicians. Raised an $8.4M seed led by Yosemite.  

Gimlet Labs — San Francisco startup improving efficiency of agentic AI workloads across heterogeneous hardware. Raised a $12M seed led by Factory.  

Graph AI — Pleasanton, CA AI pharmacovigilance tools to monitor adverse drug events. Raised a $3M seed led by Bessemer Venture Partners.  

Hyro — New York AI agents that handle patient interactions are often managed by call centers. Secured $45M from Healthier Capital, Norwest, Define, and others (total $95M).  

Keycard — SF identity and access control stack for AI agents. Announced $38M across seed (a16z, boldstart) and Series A (Acrew Capital).  

LangChain — SF tools for building and deploying AI agents. Raised a $125M Series B at a $1.25B valuation led by IVP with multiple strategic investors.  

Luster — Indianapolis AI “pre-call” coaching for customer-facing teams. Raised a $3M seed co-led by High Alpha and Ivy Ventures.  

Magic — New York platform unifying reservation and purchase data so restaurants can personalize service. Closed a $10M seed led by Lerer Hippeau with notable industry investors.  

Moonshot AI — New York/Tel Aviv startup that auto-tests, redesigns, and optimizes online stores with gen-AI. Raised a $10M seed led by Mighty Capital.  

MythWorx — Dallas developer of low-power AI models that mimic human reasoning. Closed a $5M seed at a $100M+ valuation, led by Eagle funds.  

Natural — San Francisco payment infrastructure enabling AI agents to execute transactions autonomously. Raised a $9.8M seed co-led by Abstract and Human Capital.  

Nexos.ai — Vilnius control layer for secure enterprise AI adoption between employees and AI systems. Raised a $34.8M Series A at a $348.2M valuation, co-led by Index and Evantic.  

OpenEvidence — Cambridge, MA AI medical reference for clinicians. Raised $200M at a $6B valuation, led by GV with top-tier firms.  

Paygentic — SF flexible billing and payments infrastructure for AI-native/agent-driven products. Closed a $2M pre-seed led by MiddleGame Ventures.  

Sage Care — Palo Alto AI agents for patient navigation and scheduling. Raised $20M led by Yosemite with GC and others.  

Second Nature — Tel Aviv platform that trains sales/service teams via AI roleplays. Raised a $22M Series B led by Sienna VC with Zoom participating.  

SenseNet — Vancouver AI cameras, gas sensors, and satellites for early wildfire detection. Closed a $14M Series A led by Stormbreaker.  

Serval — San Francisco agent-based AI to automate IT service management. Raised a $47M Series A led by Redpoint with First Round and General Catalyst.  

Sesame — San Francisco conversational AI plus lightweight smart glasses. Raised a $250M Series B led by Sequoia, with Spark also participating.  

Streetbeat — Palo Alto AI to automate investing, risk management, and portfolio analysis for wealth managers and retail. Raised a $15M Series A led by CDP Venture Capital (total $25M).  

Strella — New York AI moderator that conducts customer research for teams. Closed a $14M Series A led by Bessemer with Decibel and others.  

Suno — Cambridge, MA AI music generation platform. In talks to raise $100M+ at a $2B+ valuation.  

Tensormesh — San Francisco infra that boosts inference by reusing KV caches across model queries. Raised a $4.5M seed led by Laude Ventures.  

UnifyApps — New York agentic automation that connects enterprise systems (e.g., claims processing, HR). Raised a $50M Series B at a $250M+ valuation led by WestBridge, with ICONIQ.  

Uniphore — Palo Alto enterprise AI for workflow and data operations automation. Raised a $260M Series F at a $2.5B pre-money with strategic participation from NVIDIA, AMD, Snowflake, and Databricks.  

VitVio — Boston AI for tracking surgical activity and automating OR coordination. Raised an $8M seed led by Bek Ventures (total $10M).  

Wonder Studios — London studio using AI to produce commercial, partnered, and original entertainment. Closed a $12M seed led by Atomico with LocalGlobe, Blackbird, and Adobe Ventures.

Thank you

Thank you to Bob Stefanski for contributing articles during the week.

That's all from last week,

Doug Neal

#ArtificialIntelligence #TechInvesting #AIBubble #VentureCapital #Innovation #AIRevolutio #AIreport eLab Ventures