OpenAI is positioning ChatGPT to become a major advertising platform, with Chief Financial Officer Sarah Friar describing the chatbot as what might result “if Google and Meta had a baby” as the artificial intelligence company looks for new ways to turn its enormous user base into sustainable revenue.

Friar made the comparison Tuesday at Goldman Sachs’ Communacopia + Technology Conference in San Francisco, outlining a strategy that could transform ChatGPT from an AI assistant supported largely by subscriptions and enterprise contracts into a major advertising business.

She said OpenAI’s advertising operation is already generating cash at a rate that would amount to about $1 billion in annual revenue, only seven months after ads were introduced on the chatbot.

Friar attributed the “baby” analogy to Fidji Simo, the former OpenAI executive who previously oversaw the company’s commercial operations. The strategy rests on combining two advantages that have historically powered the digital advertising businesses of Google and Meta.

Google captures users at moments when they have a specific need or purchasing intention. Meta, meanwhile, has accumulated extensive information about users and their interests through its social platforms.

Friar said that ChatGPT could combine both characteristics because users often explicitly describe what they want while the system can retain context about their preferences.

“It’s high intent search, you’re telling a lot, but with the context it’s not quite people like me, it’s me, because it has all the memory of me,” Friar said.

That combination could give OpenAI a potentially powerful advertising proposition: an AI system that understands not only what a consumer is searching for, but also the circumstances and preferences surrounding the request.

OpenAI currently places basic advertisements beneath ChatGPT responses for users on its free and Go subscription tiers. Friar said the company is working toward more sophisticated formats that could be tailored to the conversational nature of AI.

She said OpenAI has already seen early indications of what a more AI-native advertising format could look like as the company develops the underlying technology and expands availability.

“This is all before we’ve really launched a format that feels truly endemic to AI,” Friar said. “That’s what I get super excited about.”

The advertising strategy underpins a major change in OpenAI’s position from only a few years ago.

In 2024, Chief Executive Sam Altman said he found the idea of combining advertising with AI “uniquely unsettling.” OpenAI is now moving in the opposite direction as it seeks additional revenue streams to support the enormous cost of operating and developing frontier AI models.

That cost structure is central to the company’s commercial challenge.

ChatGPT has grown to roughly 1 billion users, giving OpenAI one of the largest consumer audiences in the technology industry. But serving those users requires substantial computing resources, while the company is simultaneously spending heavily on data centers, AI inference, and model research.

Advertising could provide a way to monetize users who do not pay for premium subscriptions while allowing OpenAI to preserve a free tier that expands the potential reach of ChatGPT. The approach also creates a potentially important distinction between AI advertising and conventional search advertising.

A traditional search query may provide advertisers with information about what a user is looking for at a particular moment. A conversational AI system could potentially understand a much longer sequence of interactions, giving it a richer picture of the user’s objectives.

That could make advertising more relevant and potentially more valuable, but it also creates significant privacy and trust questions. The more personal context an AI system uses to target commercial messages, the more sensitive the boundary becomes between useful personalization and intrusive advertising.

OpenAI will also have to prevent advertising from compromising the perceived independence of ChatGPT’s answers. Users may be less willing to trust recommendations if they believe commercial relationships influence what the AI suggests.

The company is developing its advertising strategy as it prepares for potentially greater scrutiny from public-market investors. OpenAI and rival Anthropic both filed confidential S-1 registration statements in June as they prepare for closely watched initial public offerings. Anthropic has been more openly critical of advertising inside AI assistants, creating a potentially important difference in the business models the two companies may present to investors.

OpenAI has not set a specific IPO date, although Friar has previously indicated that a public listing could take place by the end of 2027.

The company was valued at about $852 billion in its March financing round, which raised $122 billion. That valuation places substantial expectations on OpenAI to convert its technological lead and massive user base into durable cash flows.

Advertising could become an important part of that equation.

For Friar, who previously served as Square’s CFO during its 2015 IPO and was CEO of Nextdoor when the company went public through a SPAC in 2021, the challenge is to demonstrate that OpenAI’s extraordinary AI spending can eventually produce the economics expected of a technology company approaching the public markets.

ChatGPT’s consumer business, enterprise products and advertising operation will all have to grow rapidly enough to offset the cost of running more capable models and building the infrastructure required to support them.

OpenAI’s bet is that the same conversational context that makes ChatGPT useful can also make it unusually valuable to advertisers. If that model works, the company could establish a new category of advertising built around AI-mediated decisions rather than traditional search results or social feeds.

The financial opportunity is substantial, but so are the risks. OpenAI must prove that advertising can scale without eroding user trust, compromising the quality of AI responses, or creating privacy concerns around the enormous amount of contextual information ChatGPT can accumulate.

Therefore, the “Google and Meta” comparison captures more than OpenAI’s advertising ambition. It describes the company’s attempt to combine two of the most valuable properties in digital advertising with a new interface in which consumers tell an AI system exactly what they want.

Shares of Taiwan-based electronics manufacturer Wistron fell more than 6% on Tuesday after the Nvidia supplier priced a $1.47 billion global depositary receipt offering at a discount to its existing share price to finance raw-material purchases and support its expansion in artificial intelligence infrastructure.

Wistron said Monday that it had priced 25 million global depositary receipts at $58.88 each. The offering represents 250 million newly issued common shares, with each GDR representing 10 Taiwan-listed shares.

The new shares were priced at about NT$186.24 each, a roughly 5.5% discount to Wistron’s NT$197 closing price on Monday.

The issuance represents about 7.29% of Wistron’s shares outstanding before the offering, creating immediate dilution for existing shareholders and helping explain the sharp decline in the stock.

Wistron expects the new GDRs to be issued Thursday. The company said proceeds will be used primarily to purchase raw materials in foreign currencies.

The fundraising comes as Wistron is rapidly expanding its AI-server business, which has become an important growth driver for Taiwan’s electronics manufacturing sector.

Wistron approved additional capital expenditure last month, including NT$10.5 billion for facilities in Taiwan and a combined $53 million for two U.S. subsidiaries. The investments are intended to increase capacity for future AI-related business.

The company is also accelerating its manufacturing presence in the United States.

In July, Wistron opened its first U.S. manufacturing facility, a $700 million AI-server plant in Fort Worth, Texas. The facility currently manufactures Nvidia’s GB300 Grace Blackwell Ultra systems and is expected to expand production to Nvidia’s next-generation Vera Rubin platform.

The Texas investment places Wistron closer to one of the most important trends reshaping the electronics supply chain: the geographic expansion of AI infrastructure manufacturing.

Demand for AI servers has driven substantial investment by cloud-service providers and technology companies, creating opportunities for contract manufacturers such as Wistron, which assemble high-performance computing systems incorporating Nvidia processors, networking equipment and other components.

That growth, however, requires substantial working capital.

AI servers are significantly more complex and expensive than conventional computing systems, increasing the amount of capital manufacturers need to finance components and inventories. Wistron’s decision to earmark the GDR proceeds for raw-material purchases indicates that working-capital requirements are rising alongside production.

The foreign-currency component of the funding is also significant because Wistron sources materials globally and sells into international technology supply chains. Raising dollars through GDRs gives the company a direct pool of foreign-currency funding for purchases, potentially reducing some currency-mismatch risk.

For investors, the trade-off is that the offering strengthens Wistron’s balance sheet and provides capital to pursue rapidly expanding AI demand, but the discounted issuance increases the number of shares and dilutes existing holders.

Wistron’s stock had already gained about 23% this year before Tuesday’s decline, meaning investors had priced in a substantial amount of optimism surrounding its AI exposure.

The company reported NT$895.4 billion in revenue for the second quarter and NT$14.8 billion in profit after tax.

The scale of the revenue base, combined with new U.S. and Taiwan capacity, suggests Wistron is positioning itself for continued growth in AI infrastructure rather than treating the current demand surge as a short-term cycle.

However, there is a growing shareholder concern about whether the additional capacity and working capital will generate enough incremental earnings to outweigh the dilution from the GDR issue. Analysts say that will depend heavily on the durability of AI-server demand, Wistron’s ability to secure additional Nvidia-related orders, utilization rates at its new U.S. facility and the margins it can earn on increasingly sophisticated AI systems.

Wistron’s financing therefore illustrates a broader feature of the current AI hardware boom: suppliers are having to raise substantial amounts of capital to keep pace with demand before the resulting production expansion fully translates into earnings.

The sharp share-price reaction shows that investors are willing to fund that expansion, but at a price.

Billionaire entrepreneur says communities should secure enforceable limits, financial guarantees and independent oversight before data-center developers break ground

Mark Cuban is urging local governments to negotiate aggressively with data-center developers before approving new projects, noting that communities have their greatest bargaining power before construction begins.

In a series of posts on X on Monday, the billionaire technology entrepreneur laid out a 10-part framework for towns considering their first data center. His central message was that major commitments from developers should not be left to promises or broad assurances.

Instead, Cuban said every significant undertaking should have measurable limits, independent oversight, financial backing and a practical remedy if the developer fails to meet its obligations.

A data center can deliver construction jobs, expand a local tax base and bring investment in roads, power infrastructure and other services. But the facilities can also impose long-term demands on electricity, water and public infrastructure while affecting nearby residents for decades.

Cuban advised communities to hire their own lawyers and technical experts rather than relying solely on information supplied by developers. Local officials should also evaluate the project’s full planned build-out, rather than negotiating around only the first phase of construction.

Among the safeguards he recommended are explicit limits on water consumption and noise, restrictions on backup-generation equipment and provisions requiring developers to cover infrastructure costs associated with their projects.

Water And Noise Emerge As Key Battlegrounds

Cuban placed particular emphasis on water use, cautioning communities against assuming that a data center using closed-loop cooling will have no significant water requirements.

Even systems marketed as closed-loop can consume water under certain conditions, including periods of extreme heat, he said.

Noise is another major concern. Cuban noted that conventional noise standards may not adequately address the low-frequency sound generated by cooling systems and other equipment at large data centers.

“Noise is now the most common post-approval complaint and the most active area of litigation,” Cuban said in a separate post, pointing to proposed class actions in Wisconsin and Mississippi alleging that data-center noise can travel more than a mile from facilities.

The issue has become a growing source of friction between data-center operators and communities as the rapid expansion of AI computing brings large facilities into residential and rural areas.

Cuban’s recommendations extend beyond environmental and quality-of-life concerns to the financial risks communities could face if projects are delayed, downsized, or abandoned.

He called for developers to provide decommissioning funds before construction begins, as well as guarantees from financially credible parent companies. Other protections could include cash escrow accounts or letters of credit that municipalities could draw on if developers fail to meet contractual obligations.

The objective, Cuban said, is to prevent communities from being left responsible for infrastructure or cleanup costs after a project fails to materialize.

Data-Center Backlash Grows

Cuban has repeatedly warned that communities should use their leverage while AI companies and data-center developers are competing aggressively for sites and infrastructure.

In July, he said municipalities should pursue operators that violate laws or cause documented damage, arguing that the extraordinary demand for computing infrastructure gives communities negotiating power.

“The AI companies need the data centers more than they need air,” Cuban said at the time, urging communities to impose financial consequences for documented harm.

Later that month, he told the “All-In” podcast that technological advances could eventually make some of today’s computing infrastructure obsolete. He joked that unused data centers could ultimately become “pickleball courts,” underscoring the risk of communities being left with large facilities whose economics deteriorate faster than expected.

Public opposition is already presenting a significant challenge for the industry. A Gallup poll conducted in March found that 71% of Americans opposed having a data center built where they live, compared with 53% who opposed a nearby nuclear power plant.

Environmental and community groups have focused particularly on water consumption and noise. The Environmental Health Project has warned that cooling systems and diesel backup generators can generate persistent noise audible to surrounding communities, while cooling requirements can consume substantial quantities of water in areas already facing supply constraints.

The backlash has even entered popular culture. Former NFL player Jason Kelce recently appeared in a satirical advertisement for Garage Beer and Liquid Death built around the controversy over the amount of water associated with cooling AI data centers.

For local governments, the debate is increasingly about more than whether a data center brings jobs and tax revenue. Officials are being asked to determine who pays for new power and water infrastructure, who bears the environmental costs, and who is responsible if a facility fails to deliver its promised economic benefits.

Cuban described his framework as a work in progress and invited others to contribute. But its underlying argument is that once a municipality approves a project, its negotiating leverage can diminish sharply.

For communities facing proposals from developers racing to secure land and power for the AI boom, Cuban’s advice is to negotiate the protections first and approve the project afterward.

Huawei’s long-running confrontation with the United States is entering a decisive phase, but the case is about far more than alleged sanctions violations and bank fraud.

When jury selection begins Tuesday in federal court in Brooklyn, prosecutors will present a sweeping indictment that reaches from Huawei’s dealings with Iran and North Korea to alleged theft of American trade secrets. The defense will confront not only the charges themselves, but also the broader U.S. narrative that Huawei’s global expansion was supported by deception and illicit access to technology.

The trial, expected to last about three months, comes eight years after U.S. prosecutors secretly charged the Chinese technology company. It also arrives at a moment when Huawei has become central to China’s effort to build a technology industry less dependent on the United States.

Huawei has pleaded not guilty and has denied wrongdoing.

The company faces charges including bank fraud, wire fraud, money laundering, racketeering, obstruction, and violations of U.S. sanctions against Iran. Prosecutors also accuse Huawei of conspiring to steal trade secrets from five U.S. technology companies, concealing aspects of its business in North Korea and supplying surveillance equipment to Iran that was allegedly used to monitor protesters during anti-government demonstrations in Tehran in 2009.

The case began during the first Trump administration with the 2018 arrest of Huawei Chief Financial Officer Meng Wanzhou in Vancouver. Her detention triggered a diplomatic crisis involving the United States, China and Canada and became one of the clearest symbols of the widening conflict over technology, trade and national security.

The U.S. government placed Huawei on its trade blacklist in 2019, cutting the company off from many American technologies. Washington said Huawei posed national-security risks and warned that its telecommunications equipment could be used by Beijing for espionage. Huawei has rejected those allegations.

A conviction would give U.S. officials a criminal verdict to cite in defending restrictions that have often rested on national-security assessments rather than court findings. An acquittal would not automatically remove those restrictions, but it could complicate the legal and political narrative that has supported them.

The trial is also expected to overlap with Chinese President Xi Jinping’s scheduled September 24 meeting with U.S. President Donald Trump in Washington. That timing could make the proceedings an additional source of tension as the two governments attempt to manage disputes over tariffs, advanced chips, artificial intelligence and technology exports.

Potential jurors have been asked to complete 27-page questionnaires, including questions about their views of the governments of China and Iran — an indication of how difficult it may be to separate the case from the larger geopolitical conflict.

The Iran And HSBC Allegations

The prosecution partly grew out of Reuters reporting in 2012 and 2013 on Huawei’s relationship with Skycom, a company that operated in Iran.

Reuters reported that Skycom had offered in 2010 to sell at least €1.3 million ($1.5 million) of embargoed Hewlett-Packard computer equipment to Iran’s largest mobile-phone operator. Reuters also reported that Meng had served on Skycom’s board from February 2008 to April 2009.

The U.S. indictment alleges that Huawei used Skycom to obtain restricted U.S. goods, technology and services for its Iran operations and to move money from Iran through the international banking system.

Prosecutors said an unnamed bank processed more than $100 million in U.S.-dollar transactions connected to Skycom. Sources and documents reviewed by Reuters identified that institution as HSBC.

The allegations against Meng focused in part on a presentation she gave to an HSBC executive. Prosecutors said she misrepresented Huawei’s ownership and control of Skycom, causing the bank to underestimate the risks of continuing to handle Huawei-related transactions.

Meng later acknowledged that some of her statements were false as part of an agreement with U.S. prosecutors that allowed the charges against her to be dropped. Canadian authorities released her in 2021 after nearly three years of house arrest while she fought extradition to the United States.

Her release eased one of the most volatile elements of the dispute, but it did not end the case against Huawei. The company’s trial will now test whether prosecutors can prove that the alleged misrepresentations and transactions were part of a broader corporate scheme rather than isolated actions by individual employees.

A Case That Reaches Far Beyond Iran

The indictment is broader than a conventional sanctions-evasion prosecution. U.S. prosecutors accuse Huawei of conspiring to steal trade secrets from five American technology companies, adding intellectual-property allegations to the case.

The alleged conduct spans roughly 1999 to 2020, covering more than two decades of Huawei’s international expansion. That time frame could force jurors to evaluate a large volume of corporate records, employee communications, and technical evidence while determining whether separate incidents formed a coordinated pattern.

Prosecutors have also accused Huawei of concealing aspects of its business activities in North Korea and supplying surveillance technology to Iran. Those allegations are intended to support the government’s broader claim that Huawei repeatedly used opaque corporate structures and misleading statements to operate in markets subject to U.S. restrictions.

Huawei has rejected that narrative.

“The government’s overarching narrative is demonstrably false,” a Huawei spokesperson said in a statement on Saturday.

Beijing has also criticized the prosecution, describing it as part of a broader U.S. campaign to contain China’s technological rise. That argument has become more prominent as Huawei has moved from being primarily a telecommunications-equipment maker to becoming one of the most important companies in China’s effort to develop domestic capabilities in semiconductors, smartphones, cloud computing and artificial intelligence.

The trial matters more now than it did when the charges were filed because Huawei has become a strategic test of whether China can withstand U.S. technology restrictions.

The 2019 blacklist sharply limited Huawei’s access to American components, software and semiconductor technology. The restrictions damaged its smartphone business and disrupted parts of its supply chain, but they also accelerated Huawei’s efforts to develop domestic alternatives.

Huawei has since expanded its role in China’s technology ecosystem. Its work on advanced chips has made it a symbol of Beijing’s push for semiconductor self-sufficiency and a target of continued U.S. efforts to restrict China’s access to high-end computing technology.

That creates a paradox at the center of the case. U.S. restrictions were designed to slow Huawei’s access to advanced technology, but they also helped turn the company into a national project for China. Huawei’s survival has allowed Beijing to argue that American controls can encourage Chinese companies to replace foreign suppliers rather than remain dependent on them.

The prosecution is part of Washington’s broader effort to prevent sensitive technology from reaching companies that could support China’s military and strategic capabilities. The trial therefore sits at the intersection of criminal law, export controls, industrial policy and national security.

The Consequences Could Extend Beyond Huawei

If Huawei is convicted, legal experts have noted, U.S. prosecutors are expected to seek forfeiture of property or proceeds allegedly obtained through illegal activity. The company could also face additional commercial and reputational damage, particularly in countries that have tried to balance access to Huawei equipment against pressure from Washington.

A conviction would strengthen the U.S. government’s argument that restrictions on Huawei respond to documented criminal conduct rather than simply to commercial rivalry or geopolitical distrust. It could also encourage prosecutors to pursue similar cases against other Chinese technology companies accused of sanctions violations, export-control violations or efforts to obtain restricted American technology.

The case could influence how banks, telecommunications operators and equipment suppliers assess their exposure to Huawei. Even where governments do not impose outright bans, a conviction could increase compliance costs and make companies more cautious about doing business with Huawei-related entities.

An acquittal would not necessarily reverse U.S. export controls or national-security restrictions. Those policies can be imposed on the basis of intelligence assessments, supply-chain concerns and strategic calculations that do not require a criminal conviction. But an acquittal could give Huawei and Beijing a powerful argument that Washington overstated or failed to prove its allegations.

It could also make future prosecutions more difficult if jurors reject the government’s effort to connect years of transactions, corporate relationships and technology disputes into a single criminal conspiracy.

Huawei, however, is no longer the same company it was when Meng was arrested. It has spent years adapting to U.S. restrictions and repositioning itself as a pillar of China’s technology strategy.

Arm CEO Rene Haas believes artificial intelligence could help humanity cure cancer within his lifetime. But his prediction comes with a major qualification: the technology’s medical promise will depend on whether the semiconductor industry can supply the chips, memory, data centers and electricity needed to run powerful AI systems.

“I’ve always thought that the killer app for AI is health,” Haas said in an interview with the BBC released Tuesday. “AI is going to not only shorten the amount of time that those drugs can be invented, it’s going to shorten the amount of time that you test them.”

“I believe in our lifetime, AI will help cure cancer,” he added.

Haas’s comments capture the widening gap between AI’s ambitions and the physical infrastructure required to support them. The industry is increasingly presenting AI as a tool for solving some of humanity’s most difficult problems, from drug discovery and climate modeling to robotics and advanced manufacturing. Yet the systems needed to deliver those breakthroughs are placing unprecedented pressure on semiconductor supply chains.

Haas has led Arm, the British chip-design company whose technology is used in products ranging from smartphones to data-center processors, since February 2022. He joined the company in 2013 after spending seven years at Nvidia, where he was vice president of its computing products business.

His background gives him a direct view of the infrastructure demands created by the AI boom. Arm does not manufacture chips itself; instead, it licenses processor designs to companies that build chips for phones, servers, vehicles and other devices. As AI workloads spread across those markets, demand is rising for both high-performance data-center processors and more efficient chips capable of running AI applications at the edge.

“Right now, it’s quite constrained. We need more fabs before we can put a data center in space, I’ll tell you that much,” Haas said. “We’ll probably only put data centers in space when the biggest impediment to data centers is the cost of the data center.”

His reference to space-based data centers was partly humorous, but it underscored a serious point: the industry is still struggling to expand capacity on Earth. Semiconductor manufacturers are operating in what Haas described as an “absolutely supply-constrained environment,” and he expects shortages to persist.

AI’s Hardware Bottleneck Is Broader Than Processors

The shortage is not limited to the graphics processors and custom accelerators used to train and run AI models. One of the most important constraints is high-bandwidth memory, or HBM, which allows AI accelerators to move large volumes of data quickly.

Modern AI systems require enormous amounts of memory because their models contain billions or even trillions of parameters. As models become larger and more capable, the amount of memory needed to train and operate them increases. HBM has therefore become a critical component in the AI supply chain, and its production is concentrated among a small number of manufacturers.

But that concentration creates a vulnerability. Even if chipmakers can produce more AI accelerators, they may not be able to ship complete systems without sufficient supplies of advanced memory and packaging capacity. In many cases, the limiting factor is not the processor itself but the ability to combine the processor, memory, and networking components into a functioning AI system.

Micron Chief Operating Officer Manish Bhatia described the memory shortage as “really unprecedented” in a January interview. He said demand for HBM was consuming manufacturing capacity and contributing to shortages of memory used in more traditional products, including smartphones and personal computers.

Qualcomm CEO Cristiano Amon also warned during the company’s February earnings call that an “industry-wide memory shortage and price increases” could affect the size of the handset market during the year. The comments show that AI demand is beginning to influence markets far beyond data centers.

The pressure is also spreading to advanced packaging, the process used to connect processors and memory in high-performance systems. Packaging has become strategically important because AI accelerators cannot deliver their full performance without fast connections to memory. Expanding packaging capacity can be as difficult and time-consuming as increasing chip production itself.

Building Fabs Takes Years, While AI Demand Is Rising Now

The semiconductor industry is responding with a wave of investment. Chipmakers, memory manufacturers and governments are committing tens of billions of dollars to new fabrication plants, packaging facilities and research programs.

But supply cannot expand quickly. A new semiconductor fabrication plant can cost tens of billions of dollars and typically takes two or three years to build, followed by a lengthy process to install equipment, qualify production lines and reach high yields.

The development has created a mismatch between the speed of AI investment and the pace of industrial expansion. Technology companies can deploy new models and order additional computing capacity within months, while the factories needed to produce the required hardware may take years to complete.

The result could be periodic shortages, higher prices, and greater competition among customers. Large cloud providers and AI companies are likely to receive priority because of their purchasing power and long-term contracts, potentially leaving smaller businesses, universities and startups with less access to advanced computing.

The shortage could also encourage companies to develop smaller, more efficient models that require less computing power. Improvements in algorithms, model compression, specialized chips and software optimization may reduce the amount of hardware needed for some applications. But efficiency gains may not fully offset demand if AI use continues expanding into new industries.

Data Centers Face An Energy And Public-Acceptance Problem

Even when chips are available, AI companies need somewhere to operate them. Data centers require large amounts of electricity, cooling, land, and network infrastructure.

In the United States, more than 1,400 data centers had been built or approved for construction by the end of last year, as technology companies and infrastructure developers raced to accommodate generative AI and other compute-intensive applications.

The expansion is increasingly generating opposition from local communities concerned about electricity consumption, water use, noise, land development and environmental effects. A Gallup survey of 1,000 U.S. adults published in May found that seven in 10 respondents opposed building AI data centers in their local area, with nearly half saying they were strongly opposed.

That resistance could become a significant constraint on AI growth. Data-center developers may have access to capital and hardware but still face years of delays while securing permits, connecting to power grids, and negotiating with local governments.

Electricity availability is becoming a huge issue. Large AI facilities can require as much power as a small city, and clusters of data centers can place pressure on regional grids. Utilities may need to build new generation, transmission lines, and substations before facilities can begin operating at full capacity.

Water use is another concern, especially in regions where data centers rely on evaporative cooling. Developers are exploring alternative cooling systems, including closed-loop and liquid-cooling technologies, but those approaches can increase construction costs and complexity.

AI Could Accelerate Cancer Research, But It Cannot Eliminate Clinical Risk

For healthcare, the potential benefits of more computing power are substantial. AI systems are already being used in drug discovery, protein analysis, medical imaging, patient monitoring and the identification of potential therapeutic targets.

AI can search large biological databases, predict how proteins interact, identify patterns in medical images and help researchers prioritize compounds for laboratory testing. In principle, these tools could reduce the time and cost required to move from a biological hypothesis to a candidate treatment.

But Haas’s prediction should not be interpreted as evidence that a universal cancer cure is imminent. Cancer is not a single disease but a collection of more than 100 diseases, each involving different genetic mutations, biological mechanisms, and responses to treatment.

AI may improve the odds of finding effective therapies, but it cannot remove the fundamental uncertainty of human biology. A compound that performs well in a computer simulation or laboratory experiment may fail in animals or humans. A treatment that works for one genetic subtype of cancer may be ineffective for another.

Clinical trials remain essential. Researchers must establish that a treatment is safe, determine the appropriate dosage, measure its effectiveness, and understand its long-term risks. Regulatory review and manufacturing also take time, even when the underlying discovery process is accelerated.

AI could therefore have its greatest near-term impact not by producing a single cure for cancer, but by helping develop more precise treatments for specific cancer types. It may also improve early detection, identify patients most likely to respond to a therapy, and help doctors select combinations of existing treatments.