At a pivotal moment when the AI boom is reshaping the tech sector landscape, Goldman Sachs' top TMT trader Peter Callahan has identified 10 core observations and 10 key charts, outlining the most important structural shifts in tech investing: AI agent (silicon) traffic is overtaking human (carbon) traffic, optical components are outperforming memory chips, and open-source models are achieving superior performance at less than 8% of the cost of closed-source ones. These three comparisons reflect a deep evolution in the investment logic for AI infrastructure.

In an August 9 report, Callahan maintained a constructive view on the tech sector's performance for August and beyond, citing cleaner positioning, easing technical pressures, relatively reasonable valuations, and strong fundamentals and earnings. Last week, the Nasdaq 100 jumped 5% in a single week, with the VIX closing near 14, hitting a new low for the year. Discussions on the Fed's policy path have taken a backseat following the nonfarm payrolls data.

At the industry level, the latest earnings season has further widened the sentiment divergence between data infrastructure and development tool companies on one side, and traditional SaaS application vendors on the other. The former are shifting from an AI "headwind" narrative to an AI "tailwind" one, while whether the latter can keep pace with this narrative shift will be a key market focus. Meanwhile, persistent GPU supply tightness, non-human network traffic surpassing 50%, and rapidly expanding corporate AI budgets all point to a rapidly forming "inference economy."

Three Major Structural Themes: Agents, Optics, and Open Source

Agents are replacing humans as the internet's main characters. During its earnings call, Cloudflare reported that non-human traffic has already exceeded human traffic and predicted that if current trends continue, non-human traffic will be 1,000 times that of human traffic in five years—"humans will become a rounding error on the internet." Third-party data also shows that bot traffic accounted for 53% of all web traffic in 2025, with human traffic falling below half.

Corporate AI budgets are expanding, but ROI divergence is already emerging. Booking stated that its AI investments have generated positive returns, with related costs still representing a low single-digit percentage of overall tech spending. MercadoLibre spent approximately $80 million on AI this quarter, revealing that its customer service headcount dropped from 10,000 to 7,000 even as business volume grew threefold, with 90% of customer interactions no longer requiring human intervention.

Open-source models are challenging the closed-source ecosystem at extremely low costs. Pinterest disclosed in its earnings that by fine-tuning an open-source model on proprietary data, it achieved performance surpassing closed-source third-party models, with a per-transaction cost of less than 8% of comparable closed-source models. This data poses a direct challenge to the business model of closed-source AI models and provides a powerful real-world footnote to the "open vs. closed" market debate.

Goldman Sachs' recent sales and trading reports characterize the current market as "a return to normal, not a boom," describing the slowdown in earnings growth as "a structural change, not a directional reversal." Goldman Sachs' partner Louis Miller concluded with a market adage: "The trend is your friend until it bends in the end."

1. Disney — Breakout or Breakdown?

Disney delivered a strong quarterly report, with domestic theme park attendance accelerating, but debate over the potential for earnings revisions remains. From a technical perspective, Disney's stock price has been trading in a descending channel for years and is now at a critical juncture—whether it will break upward or decline again is a core question for investors.

2. Optics vs. Storage — The Quarter's Most Significant Sector Divergence

This is one of the most frequent topics Callahan received from clients this week. In the third quarter to date, both optical stocks (blue line) and memory stocks (red line) experienced roughly a 30% pullback in July, but since August, optical stocks have outperformed memory stocks by over 20 percentage points cumulatively. Specifically, Coherent has risen for seven consecutive trading days, gaining 73% in that period. Lumentum has risen 43%, Fabrinet 36%, Corning 34%, and Ciena 24%. In contrast, NAND, DRAM, and HDD-related memory stocks have seen gains ranging from 0% to 20% over the same period, clearly lagging. This divergence reflects a market repricing of compute demand in the AI inference phase, with the explosion in optical interconnect demand exceeding previous expectations.

3. Block — Can the Stock Price Catch Up to Fundamentals?

The stock price of payments provider Block (black line) has long languished in the low $80 range, even as its gross profit (blue line) has continued to climb. The divergence between the two is becoming increasingly significant. This chart, annotated by Goldman Sachs' trader Christian DeGrasse, poses a core question: can the stock price eventually converge with fundamentals and achieve an upward correction?

4. HP — Up 40% in 40 Days, Key Resistance Ahead

HP's stock price has quietly risen about 40% in 40 trading days and is now retesting the critical resistance level near $30—a level that has acted as a "ceiling" multiple times over the past year. Whether it can break through this level effectively will be a key signal for judging the nature of this rally.

5. SaaS Observation — A Breakthrough Moment Above the 200-Day Moving Average

While market conversations are focused on finding beneficiaries of the "inference economy" (such as Snowflake, Cloudflare, Palantir, CrowdStrike, and Palo Alto Networks), Callahan notes that ServiceNow recently closed above its 200-day moving average for the first time in over a year, a signal that is hard to ignore. This aligns with his core view on SaaS/application companies: as non-quarterly reports from companies like Intuit, Adobe, and Salesforce flood in, the market will closely watch whether these companies can also build their own AI "tailwind" narratives. Notably, Atlassian has rebounded roughly 160% from its year-to-date low but is still down about 6% year-to-date—a paradoxical data point that encapsulates the current situation of the SaaS sector.

6. Nvidia — A Delayed Catch-Up Rally?

Nvidia rose 11.5% last week, its best single-week performance since May 2025. Callahan believes the catalyst path is becoming more favorable—upcoming earnings, the Vera Rubin architecture launch, and a dense industry conference season all combine to create a potential driver for Nvidia's catch-up rally. The chart shows Nvidia's relative performance against the Philadelphia Semiconductor Index (SOX), with its relative strength reasserting itself. On the GPU supply front, tightness persists. Akamai stated on its earnings call: "GPU demand remains exceptionally strong, and all our GPU capacity is fully sold out." This echoes similar statements from cloud providers like Microsoft and Amazon.

7. Spotify — The Divergence Between Positive Sentiment and Downward Revisions

Investor sentiment around Spotify's AI music releases has been quite positive, but the chart reveals a concerning reality: over the past nearly two years, the company's operating profit forecasts (orange line) have been consistently revised downward, while the stock price (white line) has held up due to sentiment-driven buying. Callahan's judgment is straightforward: "same ol' story"—the divergence between sentiment and fundamentals is the core contradiction this stock has long faced.

8. Atlassian — Two Consecutive 30% Single-Day Surges, Still Down Year-to-Date

Atlassian has seen T+1 single-day gains of over 30% in two consecutive earnings seasons, driving its stock price up roughly 160% from its year-to-date low. However, this chart reveals a highly dramatic fact: even so, Team is still in negative territory year-to-date. This data both illustrates the depth of the previous decline and serves as a reminder for investors to carefully assess margin of safety when chasing highs.

9. Uber — A GARP Narrative Emerges

Uber is once again exhibiting a pattern of "beating earnings, rallying, then pulling back," but the stock has risen about 15% over the past two weeks. Investor conversations are now beginning to include clear GARP (Growth at a Reasonable Price) and value investing logic—at about 13x forward free cash flow, with mid-to-high single-digit to double-digit revenue growth, the valuation appeal is attracting institutional attention.

10. Q2 Earnings — Information Technology Leads the S&P 500

The information technology sector led the S&P 500 in the second quarter earnings season, with 82% of its constituent companies beating earnings estimates and 72% surpassing revenue expectations. This data provides the most direct fundamental support for Callahan's constructive view on the tech sector.