OpenAI CEO Sam Altman said he underestimated how quickly artificial intelligence would reshape businesses and displace established software products, acknowledging that companies have been far slower to change their habits than the technology’s rapid development initially suggested.

In an interview with David Senra published Sunday, Altman said he had expected the release of GPT-4 in 2023 to quickly create opportunities for new software companies and force businesses to reconsider the tools they used. Instead, customers have largely continued buying from familiar vendors and using established products.

“I think it means we’ve all been too ambitious on timelines,” Altman said. “People keep doing the same things they’re doing. They keep buying from the same company. They keep sort of wanting to use their tools in the same way.”

Altman described the slower transition as “positive in many ways,” offering a more cautious assessment of AI’s economic impact than some of the predictions that accompanied the generative-AI boom.

That matters for the technology industry because AI capabilities have advanced rapidly, but technological capability and economic adoption are not the same thing. Companies must change procurement processes, train employees, integrate new systems with existing infrastructure, and become comfortable relying on AI for business-critical tasks.

Those barriers can slow disruption even when a new technology appears capable of replacing an existing product.

OpenAI and other leading AI companies have spent years arguing that capable models could make businesses more efficient, allow smaller teams to compete with established companies, and automate substantial amounts of white-collar work.

Anthropic CEO Dario Amodei has gone further, predicting that AI could eliminate as much as half of entry-level white-collar jobs within five years.

Those expectations have already affected financial markets. Software-as-a-service companies came under heavy pressure in early 2026 as investors began questioning whether AI systems could reduce demand for traditional enterprise applications.

The selloff, dubbed the “SaaSpocalypse,” hit companies including Salesforce, Atlassian and Asana as investors considered the possibility that businesses could increasingly use AI tools to create customized software rather than purchasing standardized applications.

But Altman’s latest comments suggest that the disruption may take longer to materialize at scale. He compared the situation with the transition from physical video rental to streaming. Customers continued visiting Blockbuster even after Netflix had begun offering DVD rentals by mail, illustrating how established habits can survive even when a more convenient technology is already available.

“It was amazing to me that people still went to Blockbuster,” Altman said. “That is an example that has stuck in my head of like force of habit, and the way people do things and changing behavior is just much harder than the tech nerds realize.”

The comparison points to a central challenge for AI companies. Developing a model that can perform a task is only the first step. Convincing millions of people and thousands of companies to change how they perform that task can take considerably longer.

Enterprise customers, in particular, have reasons to move cautiously. Software is often deeply embedded in corporate workflows, data systems and compliance processes. Replacing an established application with an AI-based alternative can create operational and security risks even when the new system is technically more capable.

There is also an economic question about who captures the benefits of AI. A company may use AI to perform a task more efficiently without abandoning the software vendor that already provides its broader workflow. AI could therefore initially increase the productivity of existing products rather than immediately destroy them.

Altman acknowledged that this inertia exists even inside his own working habits.

He said he still manually works through his email inbox even though OpenAI’s Codex could automate more of the process. That example illustrates the gap between what AI can theoretically automate and what users actually choose to delegate. People may continue performing tasks themselves because they prefer existing routines, want to retain control, or simply have not developed new habits around AI.

The slower adoption cycle complicates some of the more aggressive assumptions embedded in AI valuations for investors. The technology may ultimately transform software, employment and corporate operations, but the timing of that transformation is increasingly difficult to predict.

For AI companies, the implication is equally significant. Technical progress alone may not be enough to produce rapid economic disruption. Distribution, integration, trust and changes in user behavior could determine how quickly AI moves from an impressive capability into a replacement for established products.

Altman’s comments therefore amount to a reassessment of the timeline rather than a retreat from the broader AI thesis. OpenAI still expects increasingly capable models to change how people and companies work. But the experience since GPT-4 has shown that technological disruption can move at two different speeds: AI capabilities can advance rapidly while the institutions and people expected to use them change much more slowly.

Nvidia’s near-monopoly over advanced artificial intelligence chips is coming under increasing pressure as OpenAI and other major technology companies develop custom semiconductors designed to reduce their reliance on the chip giant, analysts told CNBC.

OpenAI unveiled its first AI chip, Jalapeño, on Tuesday and said initial testing showed “industry-leading speed and efficiency.” Developed in partnership with Broadcom, the chip is designed primarily for AI inference, the process of running trained models to generate responses and perform tasks for users.

The development has become of interest because inference is becoming one of the fastest-growing sources of AI computing demand. As companies deploy AI agents and models to millions of users, the amount of computing required to operate those systems after training can become enormous.

Nvidia remains the dominant supplier of AI accelerators, with its GPUs powering much of the training and inference infrastructure used by leading AI companies and cloud providers. Its CUDA software ecosystem has also created a significant barrier to switching because developers have built years of tools and applications around Nvidia’s architecture.

But the emergence of custom chips threatens to weaken that advantage, particularly among the largest technology companies that have the scale and engineering resources to design their own silicon.

Adrien Sanchez, a technology analyst at Yole Group, said Jalapeño demonstrated that a chip designed by a major cloud or AI company can now compete with Nvidia’s Blackwell-class GPUs on inference efficiency.

“Nvidia still owns the vast majority” of AI compute and retains strong software ecosystem lock-in through CUDA, Sanchez said. But OpenAI’s chip represents a “threat to Nvidia’s inference margins, which is the field growing the most at the moment.”

Nvidia’s biggest vulnerability may not be an immediate loss of its overall AI chip leadership, but pressure on the economics of individual workloads.

OpenAI is one of Nvidia’s largest customers, buying huge volumes of GPUs to train and operate its AI models. If the company can shift a meaningful portion of inference workloads to its own chips, it could reduce its dependence on Nvidia and gain greater control over computing costs.

OpenAI said Jalapeño would enable faster responses, more responsive AI agents and more reliable access as demand increases. The chip is expected to be deployed in OpenAI’s infrastructure by the end of the year, and the company said it is already developing second- and third-generation versions.

That creates the possibility of a broader strategic shift. Instead of relying almost entirely on merchant GPUs, OpenAI could eventually operate a mixed infrastructure in which Nvidia hardware is used for workloads where its flexibility and performance provide the greatest advantage, while custom accelerators handle predictable, high-volume inference tasks.

Alexander Harrowell, senior principal analyst at Omdia, described Jalapeño as an “impressive achievement,” particularly in efficiency.

“In a large-scale deployment, this would save power, cooling, and power distribution infrastructure, and contribute a lot to their unit economics,” he said.

The savings could be substantial at the scale of OpenAI’s infrastructure. AI data centers consume enormous amounts of electricity, and the cost of cooling and distributing that power adds significantly to the cost of operating large clusters. A more efficient accelerator can therefore improve economics even if its headline computing performance is not dramatically higher.

Still, Jalapeño does not mean Nvidia’s dominance is about to disappear.

Fion Chiu, an analyst at TrendForce, said OpenAI’s custom chip could reduce its reliance on Nvidia for inference over time, but Nvidia GPUs are likely to remain important for computationally intensive workloads such as large-scale model training and frontier AI.

Nvidia retains advantages in programmability, performance, its software ecosystem, and the ability to support a broad range of workloads, Chiu said.

Benchmark comparisons also require caution.

Research firm SemiAnalysis tested Jalapeño after visiting OpenAI’s facilities and found that it delivered better performance per watt than Nvidia’s Blackwell architecture in nearly all of the scenarios it tested.

But SemiAnalysis said the comparison was “somewhat incomplete and unfair” because Jalapeño uses newer HBM4 memory. Nvidia’s forthcoming Rubin platform also uses HBM4, making Rubin a more appropriate comparison.

“Jalapeño is really competing against chips like Rubin that also use HBM4,” SemiAnalysis analysts said.

Nvidia’s Rubin systems are already beginning to ship to customers, while OpenAI still has engineering samples of Jalapeño, highlighting another challenge for OpenAI: bringing its custom architecture into large-scale production and deployment.

OpenAI is also far from alone in pursuing custom silicon.

Google has developed its own tensor processing units for AI training and inference, while Meta has committed to deploying custom AI chips using Broadcom technology. Amazon has developed its Trainium family of AI accelerators, with Anthropic committing to more than $100 billion of spending on AWS technology over the next decade, including current and future generations of Trainium.

The shift is becoming notable because hyperscalers account for a substantial share of global AI infrastructure investment.

Omdia expects custom application-specific integrated circuits, or ASICs, to surpass GPUs in unit volume by 2028, although GPUs are likely to remain ahead in revenue because they are substantially more expensive.

“This is the biggest competitive threat to NVIDIA,” Harrowell said, arguing that about half of AI infrastructure capital expenditure comes from hyperscale cloud providers that already have custom-chip programmes or have the resources to develop them.

The economics explain why the hyperscalers are willing to invest heavily in semiconductor design. Building an AI chip requires enormous upfront engineering costs, but at sufficient scale the savings from owning the architecture can outweigh the cost of development.

Custom silicon can also be optimized for a company’s particular models and workloads rather than being designed to serve the broadest possible customer base. That could gradually erode one of Nvidia’s traditional advantages. Nvidia sells general-purpose accelerated computing platforms that can support a wide range of AI workloads. Hyperscalers, by contrast, can design chips around their own software stacks, models and data-center architectures.

The competitive field is widening beyond the major technology companies as well. Startups including Cerebras, SambaNova, D-Matrix, Etched and Fractile are developing specialized AI processors targeting different parts of the AI computing market.

For Nvidia, however, the biggest issue may be the changing relationship with its largest customers.

OpenAI has been one of the biggest single consumers of Nvidia GPUs. If it succeeds in deploying Jalapeño at scale, the company could gain bargaining power over future Nvidia purchases while simultaneously lowering its dependence on the supplier.

Sanchez said Jalapeño “raises the stakes for Nvidia’s largest customer relationship specifically.” That does not necessarily mean OpenAI will stop buying Nvidia GPUs. More likely, the AI industry is moving toward a heterogeneous computing model in which Nvidia GPUs, custom ASICs and other accelerators coexist.

Nvidia’s challenge will be maintaining its technological lead and software advantage while its largest customers increasingly have an economic incentive to develop alternatives. The bigger change is that AI chip competition is shifting from a market dominated by a single merchant-chip supplier toward one in which the largest AI companies increasingly control part of their own semiconductor stack.

Nvidia’s second-quarter earnings on Wednesday are shaping up as one of the most closely watched events in global markets, with options traders pricing in a potential $280 billion swing in the chipmaker’s market value as investors look for evidence that demand for artificial intelligence infrastructure remains strong.

Options on Nvidia are pricing in a 5.4% move in either direction for Thursday’s session, following the company’s results. That is smaller than the 6.5% move implied ahead of its May earnings report and well below Nvidia’s average post-earnings move of 7.4% over the past 12 quarters, according to analytics firm Option Research & Technology Services, or ORATS.

At Nvidia’s current valuation, a 5.4% move represents roughly $280 billion in market capitalization. That amount exceeds the entire market value of about 90% of companies in the S&P 500, underscoring the scale of the potential reaction to a single earnings report.

“That shows some complacency for Nvidia, and it means it’s getting more predictable,” said Matt Amberson, founder of ORATS.

The subdued options pricing also suggests investors are less convinced Nvidia will deliver the kind of earnings surprises that repeatedly produced double-digit stock moves during the early stages of the AI boom.

Chris Murphy, co-head of derivatives strategy at Susquehanna, said the market has become accustomed to Nvidia’s results.

“I think the beginning of the AI era when Nvidia was surprising everybody with the huge earnings beats and 10, 15, 20 percent moves, that’s kind of over,” Murphy said. “There’s just not a huge view that they’re going to catch everybody off-guard with some giant beat and the stock’s going to really rally.”

Analysts note that it does not mean expectations are low. Nvidia remains one of the most important companies in the global technology industry and the dominant supplier of advanced chips used to train and run many of the world’s leading AI systems. The company therefore occupies a critical position in a much larger investment cycle involving hyperscalers, cloud providers, governments, semiconductor manufacturers and data-center operators.

But investors are expected to look beyond Nvidia’s headline earnings. Revenue guidance, demand for its AI accelerators, gross margins and indications from major cloud customers about future capital expenditure are likely to be closely scrutinized.

Nvidia’s ability to sustain rapid growth is largely tied to whether its biggest customers continue spending enormous sums on AI infrastructure. The question has become more important as the market starts demanding evidence that the hundreds of billions of dollars being committed to AI infrastructure will eventually generate adequate returns.

“Return on investment from the hyperscalers is really important,” said Will Sterling, chief investment officer at TritonPoint Wealth. “That will dictate whether or not they continue to invest with their capex. If that happens, then I think that’ll be beneficial from a risk-on perspective in the entire ecosystem.”

Nvidia has recently partnered with six major financial institutions on financing platforms targeting more than $500 billion for AI infrastructure, highlighting the enormous amount of capital required to build data centers capable of supporting expanding AI workloads.

The scale of that investment has also made Nvidia’s earnings spectacular to markets beyond the semiconductor sector. This means that if Nvidia reports strong demand and raises its outlook, investors could interpret that as evidence that hyperscalers remain committed to expanding AI capacity. Such a result could support other semiconductor and infrastructure stocks while easing some concerns about whether AI spending has become excessive.

Analysts note that a weaker outlook could have the opposite effect, particularly given the increasing scrutiny of AI-related capital expenditure.

Nvidia’s shares have already been under pressure. The stock fell for a seventh consecutive session on Monday, although it remains up 11.7% this year. That compares with an 11.8% gain for the S&P 500 and a 61% advance in the Philadelphia Semiconductor Index.

The divergence reveals that Nvidia’s valuation and performance are now being judged against expectations for the broader AI industry.

The earnings report also arrives at a difficult point for growth stocks more broadly. U.S. Treasury yields have risen sharply as investors contend with persistent inflation, higher energy prices and concerns about the government’s expanding debt burden. The 30-year Treasury yield reached a 19-year high last week and remains above 5%.

Higher long-term yields are bad for technology stocks because they increase the discount rate applied to future earnings and make bonds more competitive with equities. The rise in yields has already contributed to weakness across major U.S. stock indexes, increasing the importance of Nvidia’s results as a potential catalyst for the technology sector.

Treasury Secretary Scott Bessent has sought to ease pressure in the long-term bond market through increased Treasury buybacks. Reports that Treasury could use some of its nearly $1 trillion Treasury General Account to finance those purchases instead of relying entirely on additional issuance helped push the 30-year yield modestly lower Monday.

But yields remain elevated, leaving technology investors exposed to the broader interest-rate environment.

Federal Reserve Chair Kevin Warsh’s planned speech in Jackson Hole later this week will provide another potential catalyst for markets. Investors are likely to look for clues about the Fed’s assessment of inflation, economic growth and the path for interest rates.

Against that backdrop, Nvidia’s earnings will be interpreted not simply as a quarterly scorecard but as a test of the durability of the AI investment cycle.

The options market’s 5.4% implied move suggests investors expect a substantial reaction but not the extraordinary price swings that characterized Nvidia’s earnings during the early stages of the AI boom.

That relative calm may itself be significant.

Nvidia has become so large and so central to the AI trade that investors increasingly have detailed expectations for its growth, margins and customer demand. A result that merely meets expectations may therefore produce a smaller reaction than it would have several years ago.

The greater risk is the gap between Nvidia’s guidance and the enormous spending commitments already embedded in the AI ecosystem. Analysts believe that if hyperscalers continue raising capital expenditure, Nvidia’s demand outlook could remain strong and reinforce the case for continued AI investment. But if customers begin signaling greater caution over returns on AI infrastructure, investors could question whether the current spending cycle can maintain its pace.

That makes Nvidia’s commentary on hyperscaler spending worthy of investors’ attention.

As Murphy noted, the era when Nvidia could repeatedly surprise investors with enormous earnings beats may be fading. The market is now less interested in whether Nvidia can beat expectations by a wide margin and more focused on whether the company’s growth can justify the scale of capital being deployed across the AI ecosystem.

With roughly $280 billion of market value potentially at stake in Thursday’s trading, Nvidia’s results are expected to provide the clearest near-term signal yet on whether the AI boom is entering a more mature phase or still has room to accelerate.