The artificial intelligence boom is entering a more demanding phase. HTX Research argues in a new report that the technology itself remains early in its adoption curve, while AI equities, capital expenditure and investor expectations have already moved deep into a repricing cycle. As the market shifts from GPU shortages and model scale toward token economics, enterprise adoption and free cash flow, the report suggests investors may increasingly judge AI companies less on how much they spend and more on what those investments ultimately return.
The first phase of the AI investment cycle was largely about scarcity.
Graphics processors were scarce. High-bandwidth memory was scarce. Data-center capacity was constrained. Companies capable of supplying the infrastructure needed to train and deploy increasingly powerful models became the obvious beneficiaries.
That equation is changing.
In a new report titled The Industrialization of Intelligence and the Bubble Cycle: Token Economics, Capital Expenditure, and the Repricing of Risk-Reward Across U.S. AI Equities, HTX Research, the research arm of cryptocurrency exchange HTX, argues that technological adoption and financial-market expectations are now moving at different speeds.
The distinction matters. AI deployment is still expanding across software development, enterprise workflows and cloud infrastructure. But investors have already priced in years of future growth across parts of the semiconductor, cloud-computing and technology ecosystem.
The next question is therefore less about whether AI works and more about whether the enormous cost of building AI infrastructure can generate attractive returns on capital.
That shift is visible in the metrics investors are watching. According to HTX Research, market attention is moving away from model parameter counts and headline capital expenditure toward token-production costs, task-completion reliability, usage intensity, enterprise workflow penetration and sustainable free cash flow.
In practical terms, the market is beginning to treat AI as an industrial economy rather than simply a software story.
The scale of spending illustrates why.
J.P. Morgan Asset Management estimates that five U.S. hyperscalers will collectively spend about $697 billion in capital expenditure during 2026. HTX Research notes that capital expenditure is expected to rise from approximately 33% of operating cash flow in 2023 to roughly 93%.
At that level, the financial question becomes unavoidable: how quickly can AI-related revenue and productivity gains justify the infrastructure being built?
That does not necessarily make the AI boom a bubble. HTX Research explicitly separates the technology’s underlying growth from the speculative behavior surrounding portions of its financial architecture.
Cloud revenue is expanding. Coding agents are gaining adoption. Semiconductor demand remains strong. Enterprises are increasing their AI spending.
The potential bubble, according to the report, sits elsewhere—in external financing, data-center development, private AI-company valuations and selected high-multiple stocks whose prices already assume near-perfect execution.
That creates an increasingly important distinction for public-market investors.
A company can be a major winner from AI while simultaneously being a poor investment if its valuation already incorporates too much future success.
HTX Research applies that framework across major technology and semiconductor companies including Alphabet, Microsoft, Meta, TSMC, NVIDIA, Amazon, Oracle, Micron, AMD, Arista Networks and Vertiv. It identifies Alphabet as offering the strongest overall risk-reward profile at current prices, while emphasizing that valuation comparisons require normalized earnings and cash-flow analysis rather than headline price-to-earnings ratios alone.
The report’s broader argument also extends beyond equities.
AI is increasingly becoming a shared investment theme across asset classes, including cryptocurrency. Crypto investors who previously concentrated on digital assets can now gain exposure to themes such as AI semiconductors, technology ETFs, precious metals and other traditional financial assets through emerging multi-asset trading platforms.
HTX says its TradFi perpetuals product has surpassed $2.5 billion in cumulative trading volume as of August 2026, with more than 170 supported assets spanning U.S. equities, ETFs, gold, silver, crude oil, AI semiconductors, memory, aerospace and selected pre-IPO themes.
That development points to a broader convergence between crypto-native capital and traditional financial markets.
Stablecoins can function as the funding layer. A user holding USDT can potentially move between crypto and TradFi-linked exposure without opening a conventional brokerage account or transferring money into a separate financial ecosystem.
For trading platforms, that changes the competitive landscape.
The next battleground may not simply be lower fees, deeper liquidity or faster listings. Platforms could increasingly compete on multi-asset access, portfolio allocation, wealth-management functionality and AI-powered investment tools.
This convergence is particularly significant as investors begin treating assets such as NVIDIA shares, gold, ETFs and cryptocurrencies as different expressions of the same global risk appetite.
When risk tolerance falls, capital can rotate toward defensive or traditional assets. When risk appetite returns, investors can increase exposure to crypto and high-beta technology themes. The infrastructure connecting those decisions could become strategically valuable.
There are important caveats. HTX Research is affiliated with a crypto trading platform, and its assessment of the market should therefore be read with that commercial context in mind. Its company-level conclusions are analysis rather than independent investment advice.
Still, the report identifies a genuine change in how the AI trade is being evaluated.
The first phase rewarded access to compute. The next phase may reward economic efficiency.
For companies, that means demonstrating that AI infrastructure translates into recurring enterprise demand. For investors, it means separating durable competitive advantages from valuations that already assume extraordinary growth. And for trading platforms, it means preparing for a market in which crypto, equities, commodities and AI-related assets increasingly compete for the same pool of global capital.
The AI investment cycle is not necessarily ending. It is becoming harder to value.
Market Landscape
The AI market is moving from an infrastructure-construction phase toward an AI monetization and return-on-investment phase.
Hyperscalers continue to commit hundreds of billions of dollars to data centers, networking and accelerated computing. NVIDIA remains central to the AI compute stack, while companies such as AMD, Broadcom, TSMC and Micron participate in different layers of the semiconductor supply chain.
At the application layer, Microsoft, Alphabet, Amazon and Meta are integrating generative AI into cloud, search, advertising and productivity products. Enterprise adoption is therefore broadening even as the cost of inference and infrastructure remains a major consideration.
The result is a more complex investment landscape: strong AI demand can coexist with excessive valuations.
For fintech and crypto platforms, the implication is equally important. The emergence of TradFi products inside crypto ecosystems reflects a move toward multi-asset financial platforms, where stablecoins provide a settlement mechanism and users can access multiple risk categories through one account.
Top Insights
- HTX Research argues AI equities have moved ahead of technology adoption, making cash flow, token economics and enterprise usage increasingly important to investors.
- Hyperscaler capital spending is reaching unprecedented levels, increasing pressure on Microsoft, Amazon, Alphabet and Meta to demonstrate durable returns on AI infrastructure.
- The AI bubble debate is shifting from technology to financial architecture, with valuations, data centers and private-market funding creating the largest areas of investor risk.
- Crypto and traditional markets are converging, as platforms such as HTX enable crypto users to access equities, ETFs, commodities and AI-related assets.
- Trading platforms are evolving beyond crypto execution, competing increasingly around multi-asset access, portfolio allocation and AI-powered investment infrastructure.
Get in touch with our fintech expert
