GPT-6 Set for April Launch as OpenAI Faces Internal Strife and External Pressure: Executive Infighting, Financial Strain, and Rivals Closing In
At a critical juncture in the new round of the artificial intelligence arms race, news has spread that OpenAI's next-generation flagship model, GPT-6, is set for an April launch. However, behind this highly anticipated "shot in the arm" lies the growing anxiety of OpenAI founder and CEO Sam Altman and deep internal divisions over the company's finances, strategy, and the timing of going public.
Multiple sources indicate that GPT-6, internally codenamed "Spud," completed its pre-training at the Stargate data center in Texas on March 24 and could be released as early as April. OpenAI President Greg Brockman recently confirmed the model's existence in a podcast interview, calling it a "significant change in how we think about model development" rather than an incremental improvement. Altman has told employees it is a powerful model that "can truly accelerate economic development."
GPT-6: A Technological Leap and "Super App" Ambition
Leaked technical documents show GPT-6 achieves over a 40% performance improvement in coding, reasoning, and agent tasks compared to its predecessor, GPT-5.4. Its context window has doubled from 1 million to 2 million tokens. More notably is its positioning—Brockman emphasized that GPT-6 will become a new foundational model, serving as the base for all future OpenAI models.
OpenAI is planning a product reorganization around GPT-6, aiming to create an integrated "super app" combining conversation, programming, and web browsing capabilities. This application is expected to merge ChatGPT, the Codex programming agent, and the Atlas built-in browser, with a preview version targeted for late 2026. Altman himself is a heavy user of Codex, even stating that Codex lets him "use future technology," allowing him to complete his backlog of side projects.
Concurrently, OpenAI's product division has been officially renamed "AGI Deployment," marking the first appearance of the term "AGI" in the company's official organizational structure.
Strategic Contraction: Axing Sora to Focus on the Core
To concentrate resources on betting on GPT-6, OpenAI has made significant strategic adjustments. Less than 24 hours after GPT-6 completed pre-training, the company announced the shutdown of the once-sensational video generation model Sora and canceled a $1 billion character licensing deal with Disney. Altman personally called Disney's CEO to apologize.
According to Altman's explanation, the core reason for shutting down Sora was compute allocation. Sora consumed approximately $1 million worth of compute resources daily, and the same batch of GPUs was insufficient to simultaneously support training for both a video model and the flagship large model. A deeper reason, Altman suggested, was that succeeding in the short-form video space would require a product logic and business model that would lead the company down a path he "didn't want to take"—potentially becoming just another algorithm-driven content feed factory. The Sora team has pivoted to "world simulation" research, with long-term goals指向 the robotics field.
Internal Strife: The CEO vs. CFO IPO Debate
Behind the technical "all-in" bet lies immense financial pressure and internal conflict. According to a The Information report on April 6, Altman and Chief Financial Officer Sarah Friar have a serious disagreement over the timing of OpenAI's initial public offering.
Altman wants to push for an IPO as early as the fourth quarter of 2026, partly to beat its biggest competitor, Anthropic, to the public markets. However, Friar is firmly opposed. The former Goldman Sachs banker, who led Block's (formerly Square) IPO, believes the company will be nowhere near ready for a public offering in 2026. Her calculations of a staggering $600 billion in server leasing commitments over the next five years and a cash burn rate by 2030 that could be double initial expectations clash sharply with the company's slowing revenue growth.
Multiple sources familiar with the matter say the rift between the two has become公开化. Since last August, Friar no longer reports directly to Altman but instead reports to head of applied AI Fidji Simo—an extremely rare arrangement at a major tech company. Altman has excluded Friar from several key financial decision-making场合, such as discussions with major investors about server spending. One observer described Friar's situation as: "She works for a founder with grand ambitions who wants to push spending limits as far as possible."
Although the two later issued a joint statement saying they are "completely aligned" on compute strategy, the Silicon Valley规律 is that when executives need to publicly declare unity, it often means分歧 already exists.
External Threats: Secondary Market Chill and Rivals Closing In
Amid internal纷争, OpenAI is also feeling a chill in the capital markets. According to the founder of investment firm Next Round, approximately $6 billion worth of OpenAI secondary market shares were recently seeking buyers but found almost no takers. The secondary market valuation implied is around $765 billion, a discount of roughly 10% from its peak of $850 billion.
| Company | Secondary Market Valuation/Status | vs. Peak |
|---|---|---|
| OpenAI | approximately $765 billion | ~10% discount |
| Anthropic | >50% premium to last funding round | Rising |
Note: Data synthesized from recent market information and media reports.
In stark contrast, its biggest rival Anthropic's secondary market valuation commands a premium of over 50% compared to its last funding round, showing a trend of capital flowing from OpenAI to Anthropic.
Simultaneously, the global AI model arena is witnessing a "clash of titans" in April. Besides GPT-6, at least six other heavyweight models are poised for launch:
- Anthropic's Claude Mythos: Internally assessed as a "leapfrog upgrade" with recursive self-correction capabilities, but its massive compute需求 has caused server congestion, potentially delaying its release.
- DeepSeek V4: Targeting an April release, focusing on long-term memory capabilities and深度 optimized for Huawei's Ascend chips.
- Alibaba's Tongyi Qianwen 3.6 Flagship Edition: Following the departure of key figure Lin Junyang, Alibaba urgently needs to prove its AI prowess.
- Tencent's Hunyuan 3.0: Led by Yao Shunyu, expected to debut in April, emphasizing real-world performance over benchmark scores.
- xAI's Grok 5: Backed by Elon Musk's compute resources, aiming to prove that a technical path different from OpenAI's can also succeed.
- Google's Gemini 3.2: Expected in Q2, is quietly eroding ChatGPT's market share.
The Compute War: A Shared Dilemma, Different Strategies
The困境 faced by OpenAI and Anthropic are two sides of the same "compute war."
OpenAI has adopted an aggressive "bet too big" strategy. Altman has signed compute租赁 contracts totaling approximately $665 billion, extending to 2030, with partners including Oracle, Microsoft, and Amazon AWS. Some contract terms are extremely苛刻, even requiring upfront capital to help cloud providers build data centers. This massive front-loaded investment carries significant financial risk.
Anthropic, by contrast, once appeared to "bet too small." Its relatively conservative compute reservation strategy encountered severe bottlenecks when its Claude product (especially Claude Code) experienced explosive growth. In Q1 of this year, its annualized revenue surged from $9 billion to $19 billion, but server resources couldn't meet demand, leading to frequent rate limits and outages, damaging user experience and pressuring gross margins due to临时高价租赁 of spot compute.
Currently, both companies are racing toward IPOs, targeting the same window in Q4 2026, with fundraising scales in the tens of billions of dollars. Any operational issues will be magnified under Wall Street's scrutiny ahead of a listing.
Impact Analysis: Shifting Industry Landscape and Investment Trends
OpenAI's current困境 marks the AI industry's competition entering deeper waters. The dimensions of competition have expanded from单纯的算法 and model capabilities to a comprehensive contest involving compute reserves, financial health, commercialization, and corporate governance.
For investors, a reassessment of risks for leading AI companies is needed. The chill in OpenAI's secondary market shares reflects smart money's concerns about its aggressive expansion strategy and internal governance. While Anthropic is growing rapidly, its compute瓶颈 exposes weaknesses in operational capability. AI investment is shifting from chasing "narratives" and "vision" to placing greater emphasis on sustainable business models and稳健的财务规划.
For the broader industry ecosystem, the giants'天价采购 of compute further加剧了 GPU resource紧张 and could drive up costs for all participants. Meanwhile, the active entry of Chinese players like DeepSeek, Alibaba, and Tencent预示着 a more多元 and multipolar global AI race.
Altman once predicted that it "might be only two years" before the cognitive能力 inside data centers surpasses the sum total of human society. But for OpenAI, the immediate challenge is whether it can safely navigate through this April filled with internal and external troubles. The launch of GPT-6 may determine whether this star company continues to lead the trend or becomes a唏嘘的注脚 in this史上最昂贵竞赛.
Once added, BigGo Finance appears first in Google Search Top Stories, so you get the broadest, most up-to-the-minute, and most comprehensive global financial news first.