OpenAI, Meta and SpaceXAI compete to significantly reduce corporate AI costs

Fierce competition between OpenAI, Meta and SpaceXAI in enterprise AI pricing

The competition between OpenAI, Meta Platforms ($META) and SpaceXAI (Nasdaq: SPCX) is not only reflected in model scores, but also in the pricing strategy of enterprise AI. This week, they launched their latest models, promising improved performance while lowering prices to win the favor of corporate customers.

This year, some companies have employees working on AI 24/7 and are competing to consume the most tokens. This behavior is called "token maximization." The finance department then intervened in the bill review. After vendors such as Anthropic moved from a subscription model to a pay-per-use model, employers also set usage limits.

OpenAI cuts token usage, companies question AI bills

OpenAI says GPT-5.6 can get more work done with less data processing. Tokens are small pieces of information read and generated by AI systems. Reducing token use reduces the cost of coding, research, support and other large-scale tasks.

OpenAI CEO Sam Altman addressed this point in an interview with CNBC on Thursday. He said: "GPT-5.6 is 54% more efficient at gene-coding tasks. And, you know, it performs equally or better than other best models. We really see people starting to focus on efficiency, understand their spending, and get a good return on investment. This is a big step forward for us."

Sam said the model runs faster. He also added that corporate customers are now studying their spending and returns. "Every company is thinking about AI's spending and the value it gets." he said.

Sam said OpenAI wants customers to see it as a stable and rewarding supplier. "We hope to become the best, most reliable, most trustworthy and highest return on investment partner for the company."

This issue accompanied him to Sun Valley, where he met with partners and customers. Sam said this is the first time that AI costs have become a major topic at a board meeting. "Everyone is asking what we can do to help reduce spending or add value."

SpaceXAI provides cheaper computing power, while Anthropic leases entire data centers

Elon Musk's SpaceXAI is making similar commitments with Grok 4.5. The company said the model uses tokens twice as efficiently as competing products. Companies can now compare task costs with speed and intelligence.

Anthropic competes with Elon's AI business, but it is also one of SpaceX's largest customers. Users on X recently suggested that Elon could remove Anthropic from SpaceX servers at any time, harming the company. Elon rejected the idea and wrote: "It's not my style."

He also overturned an old prediction. "I was obviously wrong about Anthropic." Elon wrote Thursday. He was referring to a September 2025 post when he said: "Victory is never within Anthropic's possibilities." The report shows that Anthropic already ranked first in the enterprise AI market share at that time.

Anthropic signed a huge infrastructure contract in May. It agreed to purchase 300 megawatts of computing power from the Colossus 1 near Memphis, Tennessee. This number covers the entire output of the data center. Musk's xAI was merged into SpaceX in February, putting the agreement under the combined SpaceXAI operations.

Anthropic will pay US$1.25 billion per month until May 2029. The contract generated approximately $40 billion in revenue for the SpaceXAI division. Google, a subsidiary of Alphabet ($GOOGL), has also booked SpaceX infrastructure. Its contract lasts until June 2029 at a monthly cost of approximately US$920 million.

Elon said Anthropic did not take reckless risks by using competitor hardware. "They are clearly current leaders in the AI field." He wrote. He called Mythos/Fable the most powerful released model and said Anthropic is likely to be ready for Mythos 2 soon. He also said he would never cut the company's connection in a way that would cause serious damage.

GPT-5.6 is delayed for several weeks due to government approval

GPT-5.6 arrived in customers 'hands several weeks later than originally scheduled because it went through government approval review. Sam called the process productive and mentioned the officials involved: Commerce Secretary Howard Lutnick, Finance Secretary Scott Besant and Commissioner Cairncross.

He said that this is the first time OpenAI has taken this path, so the company has learned how to handle the next review. Sam expects it to be smoother next time and said the government team has demonstrated strong technical knowledge. "You know, this is the first time we have experienced it. So we will learn something to do better next time, and we will start soon. I think it will be a smoother process. But I was deeply impressed by the government's technical capabilities." Sam said.