Will AI Replace Programmers? Wall Street Bets on These Vibe Coding Stocks

On June 2nd (US time), OpenAI announced updates to Codex, unveiling three major new features: Agent Plugins, which provide enterprises with skill sets for six exclusive roles; Annotations, a feature allowing users to make precise modifications to generated content exactly where needed; and Sites, which lets users transform any document into an interactive website or app preview that can be shared with colleagues and clients.

These features will be rolled out gradually on Codex, with plans to integrate Codex into ChatGPT for all subscribers. Additionally, OpenAI announced that all its frontier models and Codex now support one-click deployment on $Amazon (AMZN.US)$'s AWS platform. At the end of the event, CEO Sam Altman appeared on stage, stating that the next focus of the AI industry is proactive AI capable of running autonomously around the clock—his top priority for strategic investment over the coming year.

Altman noted that OpenAI is entering its third developmental stage, with the goal of making AI accessible to everyone. Codex now boasts over 5 million weekly active users, a more than 6x increase since the desktop app launched in February.

Meanwhile, competitor Anthropic is also strengthening its moat in "Vibe Coding." On May 28th, Anthropic released Opus 4.8 and announced a valuation of $965 billion following its Series H funding round. Previously, the WSJ reported that Anthropic disclosed to investors that revenue is expected to more than double in Q2, reaching $10.9 billion—a quarterly increase of approximately 127%. The company also projects an operating profit of $559 million for the June quarter. Back in the summer, Anthropic had provided financial data suggesting it wouldn't achieve annual profitability until at least 2028.

Both OpenAI and Anthropic are targeting "Vibe Coding" as a key battleground. Since last year, global enterprises have rushed to adopt Anthropic's suite of AI coding tools. AI Coding has become one of the most easily monetizable sectors in artificial intelligence.

According to recent analysis by SemiAnalysis, Anthropic's inference gross margin has jumped from 38% to over 70%. The reasons for this margin expansion include:

1. A significant drop in token production costs. Take the actual use case of Claude Code as an example: since agent workloads have an extremely high cache hit rate (exceeding 90%), Anthropic avoids most real-time costs for 90% of requests, yet they can still charge the standard rates. SemiAnalysis estimates that the true blended cost for Opus 4.7 on agent tasks is around $0.99 per million tokens, which is far lower than the list price of $5 per million input tokens.

2. A shift in user base toward high-end tiers. A large number of users are switching from the lower-priced Sonnet series to the premium Opus series, driving up the blended Average Selling Price (ASP). Even with significant price cuts to the Opus series, SemiAnalysis believes Anthropic's unit gross profit has actually increased, as hardware upgrades have simultaneously driven down production costs.

3. Token demand is expected to outstrip supply for the foreseeable future. Major companies capable of providing truly frontier-quality large models can price based on the economic value created by the tokens, rather than merely referencing competitive costs.

AI Coding is a crucial catalyst in the AI narrative and will benefit a broader range of companies.

Beyond model-layer players like OpenAI and Anthropic, and hardware giants like NVIDIA, Broadcom and AMD, other segments poised to benefit include:

1. Cloud Service Providers: Among the biggest beneficiaries. AI code training, inference, and Agent application deployment all run on public cloud GPUs/TPUs. OpenAI's partnership with $Microsoft (MSFT.US)$ , and Anthropic's deep cooperation with AWS and $Alphabet-A (GOOGL.US)$ Google Cloud, are driving massive traffic that translates into massive cloud-native clients.

2. Workflow Management: Once AI automatically generates code, the frequency of project iterations, code commits, and bug troubleshooting increases, driving growth in paid subscriptions for R&D management and code hosting. Key companies include $Atlassian (TEAM.US)$ , $DigitalOcean (DOCN.US)$ , and $JFrog (FROG.US)$ . Notably, DigitalOcean is up over 250% this year.

3. Data SaaS: The mass production of applications leads to surge in data storage, computing, and runtime monitoring usage. The industry standard is pay-as-you-go billing. Relevant companies include $Datadog (DDOG.US)$ , $MongoDB (MDB.US)$ , and $Snowflake (SNOW.US)$.

4. CDN (Content Delivery Network) Companies: After AI generates applications with a single click, developers deploy them to nearby edge nodes. Cloudflare Workers and Akamai Edge Cloud handle lightweight AI application hosting and local inference. AI code previews and online debugging require proximate computing power, upgrading CDNs from pure traffic distribution to edge AI computing platforms and opening up new avenues for value-added payments. The companies involved include $Cloudflare (NET.US)$, $Akamai (AKAM.US)$ , and $Fastly (FSLY.US)$ .

732K Views