AI Cash Burn

Analyzing the markets with Richie Naso, a Wall Street veteran of over 40 years and former member of the NYSE.

• DJIA 52-wk: +7.42% | YTD: +12.72% | Wkly: +0.50%
• S&P 500 52-wk: +12.81% | YTD: +16.81% | Wkly: +0.31%
• NASDAQ 52-wk: +18.72% | YTD: +22.10% | Wkly: +0.91%
• State St. SPDR S&P Retail ETF 52-wk: +3.12% | YTD: +8.34% | Wkly: +2.23%

• Rate-Cut Hopes & Inflation Data

A “tame” inflation report toward the end of the week boosted expectations that Federal Reserve would cut interest rates, which helped lift stocks. Investopedia

As rate-cut expectations rose, markets regained confidence — especially in growth and tech names. Investopedia

• Mixed Start to the Week — Risk Sentiment, Crypto, Tech Weakness

The week began with a drop in risk sentiment: the Dow fell ~0.9%, S&P ~0.5%, and Nasdaq ~0.4% on Monday. Los Angeles Times

Weakness in crypto-linked and some tech stocks contributed to Monday’s slide, reflecting broader risk-off mood. KARE 11

• Sector and Stock-Specific Drivers

Some individual stocks had outsized influence: e.g. Ulta Beauty surged after strong earnings, helping lift broader sentiment. Investopedia

However, not all parts of the market participated equally: smaller-cap and speculative names remained more volatile, reflecting lingering investor caution. AP News

TEN: VERY RARE

Technology and transportation stocks have gone up for 10 consecutive sessions together. A massive change in dynamics post FED rate cut.

✅ 1. What “Cash Burn” Means (General)

Cash burn = how quickly a company spends cash before becoming cash-flow positive.

Two forms:

Gross burn = total cash out spent per month

Net burn = operating inflow – operating outflow (usually negative for startups)

✅ 2. Why AI Companies Have Unique Cash Burn Patterns

AI businesses burn cash differently from traditional SaaS because of high:

A. Compute Costs

Cloud GPU costs (Nvidia H100, H200, TPU v5p, etc.)

Training cluster rental (AWS, Azure, GCP, CoreWeave)

Inference (serving) computing can exceed training costs at scale

B. Data Costs

Licensed datasets

Synthetic data generation

Storage

C. R&D + Personnel

ML engineers ($250k–$500k comp)

Research staff

Infrastructure teams

D. Hardware Capex (for companies buying GPUs)

Thousands of GPUs → hundreds of millions of dollars

Data center buildouts, networking fabric (InfiniBand), cooling

AI companies usually have front-loaded burn: extremely high at the beginning, then stabilizes as models are trained and productized.

✅ 3. How to Calculate Cash Burn for an AI Startup

Monthly Cash Burn = (Compute + Payroll + Data + Misc OpEx) – Revenue

Example:
Category | Monthly Cost
GPU compute: $3.2M
Engineering payroll: $1.8M
Data licensing: $500k
Storage / infra: $300k
Office + misc: $200k
Total Opex: $6.0M
Revenue: $2.0M
Net Burn: $4.0M/month

✅ 4. Compute Cost Drivers (Biggest Piece for AI)

For training:

GPU hours × per-GPU hourly cost

Example: 2,000 H100 GPUs × $4/hour × 30 days ≈ $5.8M/month

For inference:

Tokens generated × cost per 1M tokens

For large models, inference can dwarf training

✅ 5. Public AI Company Examples

(ballpark, widely reported industry numbers)

OpenAI: historically >$500M/year burn during GPT-4 era

Anthropic: ~$300M–$400M/year

Cohere / Mistral: lower burn; smaller clusters

NVIDIA buyers like Meta: capex >$10B+ for GPU infra, not “burn” but upfront investment

📈 Why Many Think AI Stocks Are Overvalued

The valuations of prominent AI-related names remain extremely lofty. For example, some analysts highlight that companies with heavy AI exposure — especially “pure-play” or high-growth-expectation firms — trade at price-to-earnings or price-to-sales ratios far above historical norms or what fundamentals justify. Wedbush

There is growing skepticism that present high valuations already “price in” a near-perfect future: continuous AI adoption, major profits from AI infrastructure, and dominant market share — in short, expectations of sustained explosive growth. If AI growth slows, or competition/margin pressure rises, overvalued stocks may drop. Wedbush

For many smaller or more speculative firms, the risk is especially high: some rely on unproven business models or future AI breakthroughs that may not translate into real earnings. FinancialContent+1

From a broader-market perspective: tech and AI stocks now represent a large share of overall stock-market value — but their share of actual earnings hasn’t kept up. That mismatch raises concerns of a valuation bubble. AInvest+1

Conclusion
AI companies typically experience much higher cash burn than traditional tech businesses because their core operations depend on expensive GPU compute, specialized talent, and large-scale data. Cash burn is front-loaded — training models, renting or buying GPU clusters, and hiring researchers all require massive capital before significant revenue arrives.

The companies that succeed reduce burn through efficient model architectures, cheaper inference, better cluster utilization, and disciplined product monetization. Ultimately, managing AI cash burn is the balance between innovation speed and financial sustainability — controlling costs without slowing down model performance or product growth.

How AI Spending is Lifting U.S. Economic Growth, and What That Means for Investors:

There is little question about the impact the artificial intelligence boom is having on S&P 500 earnings and total return.1 According to my colleagues at Zacks Investment Research, the “Magnificent 7” group of mega-cap technology stocks is on track to bring in 26% of all S&P 500 earnings this year, up from 23.2% of the total in 2024 and 11.7% in 2019. The group also made up roughly 35% of the index as of the end of the third quarter.2

But what about the impact the AI spending boom is having on the U.S. economy?

In short, it’s been significant.

By some estimates, first-half 2025 GDP growth was substantially powered by spending on data centers, information-processing equipment, and software. Excluding these categories, economic growth would have been more modest. To grasp the scale of AI’s impact, consider that the dollar value of AI data-center investment has exceeded total consumer-spending contributions to GDP in 2025. The chart below also demonstrates data centers’ contribution to total fixed private investment. It’s remarkable.

Without the lift from AI capex, economic growth may have been more modest, closer to 1.5% perhaps, in the first half. Growth is growth, but I think it’s a fair argument to frame the broader economy’s performance as more ‘steady’ than ‘booming.’ The impact of AI spending doesn’t dilute growth elsewhere, it just moves the needle in the booming direction.

Outside of the AI theme, investors can find soft patches in the economy. Retail sales in September (delayed due to the shutdown) rose just 0.2%, with noticeable pullbacks in tariff-sensitive categories such as vehicles, electronics, and clothing.5 Spending on services remained firm, however, which suggests consumers are still spending selectively and with more emphasis on value. It’s a pattern consistent with an expansion that continues, but with less broad-based momentum.

Sentiment surveys show similar nuance. The Conference Board’s confidence index fell in November to 88.7 from 95.5, while the share of households reporting plentiful job opportunities also stepped down. The University of Michigan’s survey has hovered near historical lows for months. As I wrote in a recent column, I think this is symptomatic of a “K-shaped” economy, which is relying more on high-income consumers and wealth effects than on job creation or broad wage gains. This is not a negative setup—it’s just a read on where the economy largely is today.

Does this all mean that a slowdown in AI spending would cause an economic downturn by itself? At this moment, I don’t think so. But I think it could meaningfully trim the growth rate, such that the U.S. economy would be posting more modest growth than the 2% to 3% headline rate that signals overall strength. This possibility does not suggest crouching in defensive mode and waiting for AI spending to pullback substantially—it argues for positioning in solid companies with earnings growth momentum outside of the AI trade, as Zacks Investment Management does across all portfolios.

Bottom Line for Investors

I think it’s clear that AI spending has provided a boost to headline GDP this year. When you strip out the sizable capex numbers, what you see is a modestly positive expansion versus a boom. I want to be clear—this is not a bad backdrop for long-term investors. But it does leave the cycle more sensitive to a potential shift from a single, powerful growth engine (AI capex).

I think that’s the real takeaway here. The economy is in fine shape, but it’s more dependent on one theme than usual. If AI investment keeps flowing, the expansion can keep chugging along. If it downshifts, the underlying modest growth pace may become more visible. Rather than trying to forecast when or if that happens, investors are better served maintaining balance across sectors, styles, and regions so portfolios aren’t tethered to any one story. If AI capex deflates, I could see assets rotating into under-valued areas of the market that are still seeing strong earnings growth , which is a central tenet of how Zacks’ positions portfolios.

As the economy leans more heavily on AI investment, even small shifts in momentum can reshape market performance. That makes discipline, not prediction, the real advantage.

Interest-rate policy — Investors are watching for signals from the Fed. With rate-cut expectations high, any hawkish remarks could rattle markets; dovish comments could boost them further. Federal Home Loan Bank of New York

Economic data & consumer sentiment — As we head deeper into the holiday season, spending data (retail, Black-Friday / Cyber-Monday trends) will matter. It’s a big micro/macro test. Reuters

Earnings & sector rotation — Earnings reports and sector-by-sector strength or weakness (tech, retail, consumer, small-cap) could shift flows and volatility.

Major AI companies are committed; there’s no turning back. The burn rate is a significant factor affecting companies’ earnings going forward. The stock market is coiled for a sizeable move EITHERWAY.

Remember: Every trader has a different style. Focus on your own personality. When you understand your natural tendencies, you can shape a trading approach that fits you. Doing this puts you in the best position to trade consistently and profitably.

— Richie

This content (“Content”) is produced by Richard Naso. The Content represents only the views and opinions of Mr. Naso who is compensated by TradeZero for producing it. Mr. Naso’s trading experiences and accomplishments are unique, and your trading results may vary substantially from his. TradeZero does not endorse the Content and makes no representations or warranties with respect to the accuracy of the Content or information available through any referenced or linked third party sites. The Content has been made available for informational and educational purposes only and should not be considered trading or investment advice or a recommendation as to any security. Trading securities can involve high risk and potential loss of funds. Furthermore, trading on margin is for experienced investors and traders only as the amount you may lose can be greater than your initial investment. Likewise, short selling as a securities trading strategy is extremely risky and can lead to potentially unlimited losses. Options trading is not suitable for all investors as it can involve risk that may expose investors to significant losses. Please read the Characteristics and Risk of Standardized Options, also known as the options disclosure document (ODD) at
https://www.theocc.com/Company-Information/Documents-and-Archives/Options-Disclosure-Document before deciding to engage in options trading.

TradeZero provides self-directed brokerage accounts to customers through its operating affiliates: TradeZero America, Inc., a United States broker dealer, registered with the SEC and member of the Financial Industry Regulatory Authority (FINRA) and the Securities Investor Protection Corporation (SIPC); TradeZero, Inc., a Bahamian broker dealer, registered with the Securities Commission of the Bahamas; TradeZero Canada Securities ULC, a Canadian broker dealer, member firm of Canadian Investment Regulatory Organization (CIRO) and member of the Canadian Investor Protection Fund (CIPF); and TradeZero Europe B.V., a Dutch broker dealer, authorized and regulated by the Netherlands Authority for the Financial Markets (AFM) and subject to the regulatory framework of the European Securities and Markets Authority (ESMA) under MiFID II (collectively, the “TradeZero Broker Dealers”).