Alphabet Stock Falls Nearly 4% as AI Competition and Treasury Yields Pressure Shares

Key Takeaways

• Alphabet Class A shares fell about 3.8% in the September 23 session as technology stocks came under pressure.
• Lower-priced AI models from OpenAI and Anthropic renewed questions about how much Alphabet can earn from its Gemini and cloud investments.
• Meta’s Muse assistant and rising Treasury yields added to investor concerns, although early app downloads do not establish a loss of Google Search users.

Alphabet Shares Slide as Investors Reassess AI Spending

Alphabet stock fell roughly 3.8% on September 23, as a broad decline in equities coincided with fresh competition in artificial intelligence. The retreat put the focus on a difficult question for the Google parent: can its growing AI business earn attractive returns while rival models become cheaper and alternative consumer assistants gain attention?

The move followed the release of GPT-6 Sol and Luna by OpenAI and Claude Opus 5.5 by Anthropic. Both companies positioned their new models around lower costs. At the same time, investors were looking toward Meta’s Connect event for more details about its Muse AI assistant. Higher US Treasury yields weighed on the wider stock market, making it difficult to attribute Alphabet’s entire decline to any one company announcement.

Alphabet is also spending heavily to expand AI capacity. In its second-quarter earnings call, management increased its full-year 2026 capital expenditure forecast to $195 billion–$205 billion, from a previous $180 billion–$190 billion. That commitment raises the importance of growth in Google Cloud and other AI products, as well as the cost of serving them.

Cheaper Models Put Gemini Economics Under Scrutiny

OpenAI said GPT-6 Sol costs $2 per million input tokens and $10 per million output tokens, down from $4 and $20 for its previous Sol pricing. GPT-6 Luna costs $0.10 per million input tokens and $0.50 per million output tokens, also described by OpenAI as about 50% cheaper than its earlier promotional pricing. These are API prices paid by developers, rather than a direct measure of the total cost of running an AI product.

Anthropic launched Claude Opus 5.5 on September 22. Its listed input and output prices are $4 and $20 per million tokens, compared with $5 and $25 for Opus 5. Anthropic estimates the new model costs about 40% less on typical workloads when lower cache-read costs and reduced token use are included. The distinction matters: its headline input and output token rates fell 20%, rather than 40%.

For Alphabet, the immediate issue is competitive pricing across model APIs and cloud services. Google offers its own Gemini family while also supplying the computing infrastructure that businesses use to build AI applications. If comparable models become cheaper, Google may face pressure to adjust prices or improve performance. Neither outcome alone proves that Gemini margins have declined. Lower inference costs, higher usage and a broader customer base could offset lower prices per token.

The financial test is therefore broader than a model price list. Investors will watch whether Google Cloud revenue grows fast enough to cover spending on data centres, chips, power and depreciation. Alphabet said in July that faster capacity delivery drove its higher capital expenditure range and that related infrastructure costs would put pressure on operating expenses.

Meta’s Muse Raises Questions About Search Habits

Meta’s Muse assistant adds a second competitive angle. App-market estimates put its downloads at more than 2.8 million in its first 12 days, indicating a strong launch. But downloads are an early adoption measure: they do not show how often people use the assistant, whether they continue using it, or whether they have reduced their Google searches.

Investors are interested in what task-based AI assistants might do to the conventional search journey. If users increasingly ask an assistant to research products or complete purchases, some interactions that once began with a search query could move elsewhere. The effect on advertising would depend on user retention, the kinds of tasks completed, and whether Google captures the activity through Search, Gemini, Android or other products.

Alphabet has its own counterweight. Management said its AI features were increasing Search usage, and its second-quarter earnings call described more than one billion monthly active users of AI Mode. Search advertising and YouTube also remained important sources of revenue. Those figures show that the company has distribution at scale, even as competitors test new ways for consumers to interact with AI.

Meta Connect may provide another indication of how Muse will develop, especially around product integration and monetisation. A launch-period download count, however, is too narrow to establish a lasting change in Alphabet’s search market position.

Higher Bond Yields Add to Pressure on Technology Stocks

Alphabet’s decline came as stronger US economic data pushed Treasury yields higher and stocks lower. S&P Global’s September flash purchasing managers’ index pointed to the fastest US business growth in more than five years. Such readings can reinforce expectations that borrowing costs will stay elevated, particularly when investors are already focused on inflation and Federal Reserve policy.

Rising yields can weigh on the valuations of large technology companies because future cash flows are discounted at a higher rate. The effect may be more pronounced when a company commits substantial cash to projects whose returns will arrive over several years. For Alphabet, the combination of a large infrastructure programme and greater AI pricing competition put both sides of that calculation in focus.

This market-wide backdrop limits any simple explanation for the share move. Investors were reassessing growth-stock valuations across the market while also weighing Alphabet-specific questions about AI products, Search and capital spending.

What Matters Next for Alphabet Stock

The key evidence will come from customer demand, margins and cash flow rather than model announcements alone. Alphabet’s next results can show whether Cloud growth is keeping pace with infrastructure spending, whether higher depreciation is squeezing profitability, and whether AI features continue to support Search engagement.

Changes in rival API pricing and sustained use of Muse will also be relevant. More downloads or a cheaper model could intensify competition, but neither is a direct measure of lost Alphabet revenue. Conversely, robust Search activity would not eliminate the need to demonstrate returns on a $195 billion–$205 billion capital expenditure plan.

Alphabet’s September 23 selloff reflected overlapping pressures: a weaker stock market, higher bond yields and renewed debate over the economics of AI. The company still has large Search, YouTube and Cloud businesses, but investors now have a clearer set of measures to judge whether its spending can translate into durable earnings growth.