Goldman Sachs Is Split on the AI Rally: Is Earnings Growth Peaking, or Has $1.1 Trillion in Backlog Validated the Investment?

The artificial intelligence trade is entering a more difficult phase. The central question is no longer whether companies and investors are willing to spend on AI. It is whether that spending can generate durable profits, positive cash flow, and defensible value for shareholders.

That debate has now reached Goldman Sachs, where two senior strategists have offered sharply different interpretations of the same market signals. One sees slowing earnings momentum and a growing risk that AI enthusiasm has moved ahead of fundamentals. The other believes cloud demand remains strong enough to support a continued rally.

For the blockchain and cryptocurrency industry, this disagreement is especially important. Crypto markets have repeatedly adopted narratives around infrastructure, decentralized computing, tokenized AI services, and digital assets linked to emerging technologies. If traditional technology companies struggle to convert AI demand into free cash flow, crypto projects built around similar themes will face an even higher standard of proof.

Two Interpretations of the Same AI Data

Tony Pasquariello, a senior strategist at Goldman Sachs, has argued that the latest technology earnings season may indicate a broader slowdown in S&P 500 profit growth. In his view, strong results from cloud and AI-related businesses may be masking weakness elsewhere in the technology sector.

The concern is not that AI demand has disappeared. Rather, the concern is that companies are spending aggressively on data centers, networking equipment, chips, and energy infrastructure without yet demonstrating that these investments will produce proportional earnings growth.

By contrast, London-based strategist Mark Wilson remains constructive. He points to the scale of cloud providers’ contracted demand as evidence that businesses are already committing substantial resources to AI infrastructure.

According to the figures cited in the debate:

• Google Cloud revenue increased by 103% year over year.
• Google Cloud’s backlog rose 256% to approximately $561 billion.
• AWS revenue grew 39%.
• AWS backlog increased 189% to around $578 billion.
• Combined backlog for the two cloud businesses reached roughly $1.14 trillion.

Alphabet’s official third-quarter financial results and Amazon’s investor-relations materials provide the underlying corporate reporting context for the cloud expansion.

Wilson’s argument is straightforward: large enterprises would not sign such extensive commitments unless they expected AI to become an important part of their operations. From this perspective, cloud backlog is not merely a speculative forecast. It is evidence that customers are preparing to pay for AI-related services.

Backlog Is Not the Same as Realized Profit

The weakness in that argument is that backlog does not automatically translate into immediate revenue or free cash flow.

A customer may commit to future cloud capacity, but the provider still needs to build data centers, purchase accelerators, expand power infrastructure, hire specialized staff, and manage financing costs. The commercial value of a contract depends on execution, pricing, utilization, and margins.

This is why Alphabet’s cash-flow performance has become a focal point. Despite its substantial cloud backlog, the company reportedly recorded approximately $910 million in free cash flow outflow during the quarter, compared with market expectations for roughly $3.5 billion in positive free cash flow.

That contrast captures the current AI investment debate:

Demand may be real, while returns on invested capital remain uncertain.

The same distinction matters in blockchain. A decentralized computing network may attract hardware operators, developers, and token holders, but these participants do not automatically create a sustainable business. The network must prove that customers will pay for services, that revenue can cover operating costs, and that token incentives will not become the primary source of demand.

The Real Question: Who Captures the Value?

Goldman Sachs strategist Rich Privorotsky has described the challenge as a question of “value capture.” The AI ecosystem may grow rapidly, but growth alone does not identify which companies will earn attractive profits.

Value can be distributed across multiple layers:

1. Semiconductor and hardware manufacturers
2. Cloud platforms and data-center operators
3. Energy and networking infrastructure providers
4. Enterprise software companies
5. Application developers
6. Data owners and model providers
7. Financing institutions and infrastructure funds

The same issue exists in crypto markets. A blockchain protocol may generate high transaction activity without creating meaningful value for token holders. A decentralized AI platform may attract attention without developing sustainable fee revenue. A token can become liquid and widely traded while the underlying network remains economically fragile.

This is why investors should distinguish between:

• Usage and paid usage
• Token circulation and protocol revenue
• Capital expenditure and productive investment
• User growth and customer retention
• Backlog and realized cash flow
• Market attention and long-term economic value

These distinctions are particularly relevant to AI crypto projects, where marketing narratives can move faster than measurable business performance.

Why the AI Debate Matters for Crypto Infrastructure

Blockchain and AI are increasingly discussed together, but their economic models are not identical.

Traditional cloud providers usually control the hardware, customer contracts, billing systems, and service infrastructure. Their investment decisions can be evaluated through revenue, margins, capital expenditure, and cash flow.

Decentralized networks distribute infrastructure ownership across many participants. This can create advantages such as permissionless access, geographic diversity, and potentially more flexible capacity. However, it also introduces additional risks:

• Incentive programs may attract supply before demand is established.
• Token prices can obscure weak underlying cash flows.
• Governance decisions may affect network economics.
• Service quality can vary across independent operators.
• Financing costs can rise sharply during market stress.
• Revenue may not accrue directly to token holders.

For crypto investors, the Goldman debate offers a useful framework. If the traditional AI sector is judged by return on capital and free cash flow, blockchain-based infrastructure should be evaluated by the same standards.

A project claiming to provide decentralized compute should be able to answer practical questions:

• Who are the paying customers?
• What percentage of revenue comes from organic demand rather than token incentives?
• How are operators compensated?
• Can the network maintain service quality at scale?
• Does the token have a genuine role in settlement, security, or governance?
• What happens if the token price falls by 50%?

These are not merely technical questions. They determine whether a protocol is building durable infrastructure or simply packaging a popular narrative.

Market Breadth Adds Another Warning Signal

The divergence between AI-related stocks and the broader market also complicates the bullish case. When a small group of technology companies drives index performance, headline strength can conceal weakening participation underneath.

Recent market discussion has highlighted that the S&P 500 excluding major AI-linked companies performed better than technology stocks during October. This type of divergence does not prove that an AI downturn is imminent, but it does suggest that investors are becoming more selective.

For cryptocurrency markets, narrow leadership can create additional volatility. Bitcoin and major digital assets may benefit from liquidity and institutional demand, while smaller infrastructure and AI-linked tokens can experience much sharper repricing when the market questions future growth.

This makes portfolio construction and custody discipline increasingly important. A strong narrative does not remove the operational risks of holding digital assets. Investors still need to manage private keys, transaction approvals, exchange exposure, smart-contract risk, and concentration risk.

What Crypto Investors Should Watch Next

The next phase of the AI market will likely be judged by evidence rather than promises. Several indicators deserve close attention:

1. Cloud utilization and margins

Large backlogs are encouraging, but investors need to see whether contracted capacity converts into profitable revenue. Rising demand with falling margins may indicate that infrastructure costs are absorbing much of the value.

2. Free cash flow

Cash generation will become more important as capital expenditure remains elevated. Companies that consistently spend more than they earn may need external financing, creating greater sensitivity to interest rates and market conditions.

3. Enterprise retention

AI adoption is more durable when customers renew contracts, expand usage, and integrate services into critical workflows. One-time experimentation is less valuable than recurring demand.

4. Sustainable token economics

For decentralized networks, token incentives should gradually support real usage rather than permanently substitute for it. Investors should examine fee revenue, emissions, unlock schedules, and the relationship between network activity and token demand.

5. Security and self-custody

As more projects combine AI, decentralized finance, and tokenized infrastructure, users may interact with a wider range of smart contracts and applications. Keeping long-term assets separate from experimental activity can reduce the impact of a single compromised account or malicious transaction.

A More Balanced View of the AI Rally

The two Goldman strategists may not be directly contradicting each other. Both can be correct under different time horizons.

Mark Wilson’s data suggests that AI demand is real and that major cloud customers are making significant commitments. Pasquariello’s concerns suggest that demand alone is not enough to justify unlimited spending or valuations. The market must still determine how much profit can be generated after infrastructure, energy, financing, and operating costs are included.

That is also the central lesson for blockchain investors. A growing sector can contain both genuine innovation and overvalued assets. The presence of real demand does not guarantee that every company, protocol, or token will capture economic value.

As the AI and crypto industries converge, the strongest projects will likely be those that can demonstrate measurable utility, transparent economics, and disciplined capital allocation. Investors should be cautious of narratives that rely on backlog, partnerships, or user growth without explaining how those metrics become sustainable cash flow.

For individuals managing digital assets across long-term holdings and higher-risk experiments, a hardware wallet such as OneKey can provide an additional layer of self-custody and transaction control. Its value is particularly relevant in a market where speculative themes evolve quickly and users may interact with unfamiliar blockchain applications. Secure key management does not eliminate investment risk, but it can help investors separate custody risk from the broader uncertainty surrounding AI, crypto infrastructure, and token economics.

The AI rally has not yet answered the most important question: how much value will ultimately be captured by each layer of the ecosystem? Until that becomes clearer, both traditional technology investors and crypto market participants should treat impressive demand figures as a starting point for analysis—not as proof that every investment will produce attractive returns.