AI Market Update: Q1 2026 in Review

Capital expenditure sparked concerns about stretched balance sheets and a “Saspocalypse” for legacy software providers, but the underlying demand remains historically unique.

In Q2, the investment narrative is shifting toward the return on AI, the rise of autonomous agentic systems and a high-stakes race to secure the power and memory required to move AI from the screen into the physical labor economy.

Capex supercycle and agentic AI shift

A massive concentration of capital flowing into the physical and structural foundations of AI was the overarching theme in the first quarter of 2026. Projections rapidly accelerated from US$350 billion to the US$800 billion to US$850 billion range as hyperscalers and neocloud players alike aggressively expanded.

On a call with the Investing News Network, Purpose Investments analyst Nicholas Mersch said that as mega-cap companies move toward net debt, investors are being forced to rethink free cashflow and balance sheets:

“You can see people worrying about these Mag 7 stocks as a basket of them are now well into a correction. That’s why a lot of these companies have been struggling on a year-to-date basis, and people are starting to worry about the return side of it.”

Private market investments scaled dramatically, with new entrants taking a seat at the table alongside major startups like Anthropic and xAI, both securing massive funding rounds. However, risks in this lending space, which backs roughly 20 percent of software loans, have emerged — AI agents, led by programs developed by Anthriopic, caused huge disruptions for software companies serving legal, customer service and coding markets.

Weeks of selling exacerbated overspending apprehensions during the quarter, underscored by Morgan Stanley (NYSE:MS) analysts forecasting an 8 percent default risk in the event that AI automation permanently erodes the recurring revenue assumptions used to underwrite software loans.

These events revived comparisons to the dot-com era, but Mersch stressed an important contrast: “(Back then), they overbuilt a lot of the capacity, and none of it was being utilized. But if you look at these massive data centers today, they’re being instantly lit up, and they’re always sort of undersupplied and have way too much demand.”

During this stage of the AI evolution, both agentic and reasoning-based systems demand immense resources, specifically in terms of GPUs, networking infrastructure and memory. The substantial memory requirements for agentic AI triggered a parabolic surge in stock prices for firms operating within this niche.

According to data from TradingPedia, Micron Technology (NASDAQ:MU) stood out in the semiconductor industry by achieving a 360 percent return over 12 months and leading the sector with 21 percent quarterly revenue growth.

Mersch sees earnings power shifting back toward hardware, while the ultimate software winners are still being sorted out. Broad, horizontal SaaS is facing slower growth, pricing pressure and disruption from AI agents.

He’s much less bullish on incumbent software, but still favors Google (NASDAQ:GOOGL), which overtook Apple’s (NASDAQ:AAPL) market cap in early January, citing the company’s custom silicon, dominant internal models and massive distribution platform. “They’re really monetizing AI, both in cloud and across the consumer stack,” he said.

Mersch added that he has “definitely changed (his) tone” on Microsoft (NASDAQ:MSFT) due to the company’s underinvestment in data centers, which has forced OpenAI to seek other partners instead of scaling primarily on Azure. He also noted that Microsoft lacks a viable custom silicon solution and leading in-house model.

“They’ve still got an incredible cloud business, lots of distribution models, but they have to kind of find their way back in, because they’re losing in some of the key categories where Google is clearly dominating.”

As an investment theme, Mersch sees agentic AI as an under-hyped, labor‑scale opportunity that will create new winners and crush legacy SaaS models. ARK Investment Management reinforces this sentiment, forecasting that global enterprise software revenue will scale from an estimated US$1.4 trillion today to over US$7 trillion by 2030, with revenue allocation shifting from traditional SaaS models to platform-as-a-service and infrastructure-as-a-service.

Traditional software was bound by IT budgets, whereas agentic AI, in Mersch’s view, goes after the entire global labor economy, around US$50 trillion. A much larger total addressable market creates structural headwinds for software, and creates huge upside for new agent‑native companies.

“I think that this cycle is probably one of the most innovative and disruptive forces we’ve ever seen,” he said.

Mersch acknowledged “anxiety” in the venture capital community that “everything’s already done” once agents are this capable, but said he rejects that view. “I think that’s the wrong approach, because I think there’s so much white space out there where you could build some really, really cool stuff,” he explained.

As a founder opportunity, he thinks small, agent‑augmented teams can now do what used to require large, well‑funded companies, making this “probably the best time in history” to start a business.

Robotics and the physical AI frontier

Beyond agentic AI, Mersch explained that the forthcoming technical frontier lies in world models that use real-world physics and sensory data to move AI into the physical realm of robotics. Industry milestones this quarter signaled a shift from experimental prototypes to dependable, high‑performance machinery.

In his view, this physical AI buildout makes robotics a natural next stage for the AI investment cycle, with the most compelling near‑term opportunities being in industrial and manufacturing use cases.

In its Big Ideas 2026 report, ARK Invest analysts highlight robotics as a key acceleration sector, noting that the convergence of AI agents and physical hardware is shifting the industry’s revenue focus from simple automation to “intelligent, self-evolving systems” capable of independent operation in complex environments.

The mainstreaming of intelligent hardware was further underscored by announcements at the CES even in January, where Boston Dynamics and Google DeepMind outlined plans to embed Gemini-based AI into humanoid robots, supported by new specialized robotics silicon from Qualcomm (NASDAQ:QCOM) and Intel (NASDAQ:INTC).

NVIDIA’s (NASDAQ:NVDA) GTC, Q1’s other major industry event, provided public validation of the physical AI trend, with CEO Jensen Huang unveiling the GR00T N1.7 and N2 series, foundation models specifically designed for humanoid robots to learn general skills rather than being programmed for single tasks.

Mersch, however, differentiated between these longer‑dated humanoid efforts, largely because they require richer world‑model data and more complex real‑world understanding, keeping them temporarily sidelined from broad commercial deployment. While GTC featured humanoid demos, NVIDIA has also announced major industrial partnerships with Mercedes-Benz Group (ETR:MBG,OTCPL:MBGAF), Samsung Electronics (KRX:005930) and PepsiCo (NASDAQ:PEP) to deploy agentic AI in manufacturing.

The year of the “mega IPO”

Looking ahead, Mersch identified sectors with scarce supply, such as memory and power, as the primary beneficiaries.

As reasoning and agentic AI roll out, the expert expects to see structurally higher demand for dynamic random-access memory and high‑bandwidth memory, arguing that workloads built on key-value caching will require so much memory that it could become a core bottleneck of the AI buildout.

Even after short, narrative‑driven selloffs like the one that ensued after Google’s reported algorithmic breakthrough that uses less memory, he sees the long‑term trajectory as one where tight capacity, rising spot prices and recurring AI upgrades make memory a key forward‑looking beneficiary of the capex supercycle:

“This scare around headlines is a perfect example of exactly where we are in the market right now. It’s so extremely narrative-driven. It is a little bit to do as well with the overall ecosystem, more generally with geopolitical concerns, but (it’s) really interesting in terms of the dynamics that we’re seeing in the overall marketplace being so knee-jerk based and very reactionary.”

Meanwhile, power demand projections for data centers are driving a surge in behind-the-meter energy production to bypass the political and physical limits of the public grid.

That forces a surge in natural gas turbines now, and nuclear power investment over time. Mersch sees power availability as another potential bottleneck that could eventually slow the pace of AI capex if it isn’t solved.

Companies have turned to creative solutions, including BYOP models, and are exploring nuclear options, such as small modular reactors and potential plant restarts. To address memory shortages in AI workloads, companies like Google and NVIDIA are embracing CXL technology, which allows servers in a data center to share memory.

Analysts, including Mersch, believe the latter half of 2026 will bring a transition of value from the private AI sector to public markets via “mega IPOs,” with prominent entities such as OpenAI, SpaceX, Anthropic and Cerebras positioned as leading candidates. However, Mersch warned that these companies are set to go public at very high valuations compared to earlier winners, and noted that retail enthusiasm could strongly influence trading.

For the upcoming quarter, Mersch is watching for return on AI metrics and signs of peak capex to see how much debt the market will tolerate. He also is keeping an eye on potential contagion from private credit markets as valuations in that ecosystem begin to face reality. The AI capex boom is real and utilization‑driven, but unless funding and infrastructure keep pace, stretched balance sheets and private‑credit strains could curb the upside.

Don’t forget to follow us @INN_Technology for real-time news updates!

Securities Disclosure: I, Meagen Seatter, hold no direct investment interest in any company mentioned in this article.

Editorial Disclosure: The Investing News Network does not guarantee the accuracy or thoroughness of the information reported in the interviews it conducts. The opinions expressed in these interviews do not reflect the opinions of the Investing News Network and do not constitute investment advice. All readers are encouraged to perform their own due diligence.

The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.