Business

Crystal Ball: Will the AI bubble burst or balloon in 2026?

Published

5 days ago

January 5, 2026

Welcome to our 2026 Term Sheet Crystal Ball! This week, I’ll be publishing all your 2026 predictions, from the precise to the wacky, from an open call to all Term Sheet readers that I put out in December. We got some great answers from unicorn founders, key investors, and needle-moving executives.

Our first topic is one you know, love, and perhaps fear: AI. For the last few years, every year has been declared the year of AI. But 2025 was especially undeniable: The discourse around AI truly moved from boardrooms to kitchen tables. And we got comfortable, more or less, saying that there is an AI bubble.

The question, of course, is what happens now. How big is that bubble? What areas of AI are actually bubbly? And which companies have staying power, when all is said and done? Here’s what Term Sheet’s human readers think 2026 has in store.

Note: Answers have been edited for clarity and brevity.

The AI bubble

The AI hype cycle will pop. 2026 marks the end of AI novelty purchasing. Companies will only pay for AI that increases revenue, reduces churn, or automates real work. The bar moves from impressive demos to measurable ROI. —Stevie Case, Vanta chief revenue officer

A lot of AI valuations will take a markdown, squeezing compensation for AI researchers at tier-one startups. The best AI companies will continue to succeed but maybe not at the extent that they’d thought. —Deedy Das, Menlo Ventures principal

For AI buyers, 2026 will be the year of ROI. The shine of “pure potential” is wearing off, and budgets will increasingly flow to products that prove value, not just promise it. —Meera Clark, Redpoint Ventures partner

The AI bubble debate will rage on. —Jai Das, cofounder, president and partner, Sapphire Ventures

AI will need to prove itself in the numbers. —Anthony Georgiades, Innovating Capital founder and general partner

The AI funding bubble will burst when short-term investors exit. The math is simple: $200 billion invested in a single year must produce multiple trillion-dollar companies within five years—outcomes that historically take decades. Investors are being given 48 hours to decide whether to invest tens of millions into a company. That time compression tells you everything about a potential bubble. —John Kim, Sendbird cofounder and CEO

By the end of 2026, the financial markets will begin to reckon with the realities of AI business models and capital will begin shifting away from the “spend at all costs” approaches of hyperscalers and foundation models. The receding tide will result in reduced valuations in 2027 in both public and private markets. —Nnamdi Okike, cofounder and managing partner, 645 Ventures

In 2026, the AI popular narrative will pivot hard from the “God model” view to the “menagerie of models’ view. Rather than one top lab with a single “God model” dominating, there will be a plethora of thousands to millions of specialized models, distillations, fine-tunes occupying various “ecological niches.” —Jared Quincy Davis, CEO and founder of Mithril

Vertical AI becomes the survivor class of the AI cycle. If the AI bubble deflates in 2026, the companies that survive will be vertical AI platforms with real margins, real customers, and real proprietary data. —Francisco Martin-Rayo, Helios AI cofounder and CEO

Sorry SF, those AI billboards aren’t going anywhere, but they will tell an interesting story next year. The freeway will act as a live scoreboard for the sector. With thousands of AI companies fighting for oxygen, the ones that stay visible—literally—might be the ones still standing. —Merrill Lutsky, founder and CEO, Graphite

Boring incumbents will steal the AI spotlight in 2026. The last two years were about who could build the biggest model. 2026 will be about who owns the weirdest, deepest, messiest data. —Fred Hoch, TechNexus Venture Collaborative cofounder and general partner

The application layer

Consumer AI slop backlash will lead social networks to crack down and add more guardrails around AI-created accounts and content. —Amy Wu Martin, Menlo Ventures partner

The value in AI will accrue to companies that control real workflows, real assets, and real data—not those sitting at the thin application layer. —Patrick Chun, founder and managing partner, Juxtapose

In-person connection will become a premium product category. In an AI world, real human connection will become more scarce and more valuable. Consumers and workers alike are yearning to build deep relationships. We’ll see an explosion of companies that enable richer in-person experience. —Peter Deng, general partner, Felicis

AI will earn a mixed reputation in mental-health care: hopeful, powerful—and unsettling. Its unpredictability will force a reset on where and how it’s safely used. —Liam Donohue, cofounder and managing partner, 406 Ventures

In 2026, AI products will finally cross the uncanny valley and become the new normal in many instances of human engagement. —Rudina Seseri, founder and managing partner, Glasswing Ventures

Foundation models are fine; the application layer is doomed. The juggernauts (Anthropic, OpenAI, DeepMind) are likely appropriately valued or even undervalued and will continue succeeding. The real bubble risk is in the application layer, where there are now a dozen companies doing AI-powered accounting, for example. —Kamran Ansari, Infinity Ventures venture partner and Kapital Ventures founder and managing partner

AI will dissolve the boundary between finance and operations. 2026 is no longer about models giving advice, it’s about models actually executing, within guardrails. —David Roos, Core Innovation Capital partner

Consumers will prefer AI customer service reps over humans. In 2026, a patient calling their health insurance or a traveler calling their airline will prefer to deal with an AI, instead of getting routed to a human. —Jon Keidan, founder and managing partner, Torch Capital

Agents and LLMs

2026 will mark the end of the ‘bigger is better’ era in AI. LLM and AI Agent memory will improve dramatically, allowing small, specialized models with long-term memory to replace bloated LLMs in many agentic systems. —Scott Beechuk, partner, Norwest

2026 will undoubtedly be the year of the agent. —Stefan Miedzianowski, DeepL chief scientist

In 2026, the practical applications of agentic AI will overtake the promise of prompt-based generative AI in delivering real value to B2B enterprise applications. —Richard de Silva, founder and managing partner, Lateral Investment Management

This is the year open-source models become good enough. You don’t need to be at the frontier; for selective applications, you will start seeing a larger variety of models being used. This will make running models more cost-efficient, spurring new entrants and broader AI investment across the ecosystem. —Zach Lloyd, founder and CEO, Warp

Some 90% of AI agents fail within 30 days of deployment. The companies that succeed will be those with the best human-AI collaboration frameworks. Autonomous agents lead to more human oversight, not less. —Phelim Bradley, cofounder and CEO, Prolific

See you tomorrow,

Allie Garfinkle
X: @agarfinks
Email: alexandra.garfinkle@fortune.com
Submit a deal for the Term Sheet newsletter here.

Joey Abrams curated the deals section of today’s newsletter. Subscribe here.

PRIVATE EQUITY

– Galileo Education, a portfolio company of Millpond Equity Partners, acquired Putnam Classical Academy, a Palatka, Fla.-based private school and New Generation Christian School, a Lake City, Fla.-based private school. Financial terms were not disclosed.

EXITS

– Stonepeak agreed to acquire a majority stake in Castrol, a Berkshire, U.K.-based engine oils, industrial fluids, and greases manufacturer, from BP for $10.1 billion.

– TransDigm Group agreed to acquire Stellant Systems, a Torrance, Calif.-based designer and manufacturer of radio frequency and microwave amplification products, from Arlington Capital Partners, for $960 million.

– VSE Corporation acquired Aero 3, a Bedford, N.H.-based maintenance, repair, and services provider for aircraft wheels and brakes, from GenNx360 Capital Partners, for $350 million.

– Advance Technologies System agreed to acquire Conexus, a Rome, Italy-based designer, constructor, and maintainer of power transmission and distribution infrastructure, from Mutares. Financial terms were not disclosed.

– Michelin agreed to acquire Tex-Tech Industries, a Kernersville, N.C.-based designer and manufacturer of textiles and fabrics for the aerospace, space, and defense & security industries, from Arlington Capital Partners. Financial terms were not disclosed.

OTHERS

– Legence Corp. acquired The Bowes Group, a Beltsville, M.D.-based provider of mechanical, plumbing, and process system solutions, for $325 million.

– MetLife Investment Management acquired PineBridge Investments, a New York City-based asset manager. Financial terms were not disclosed.

Source link

Business

Trump calls for one-year cap on credit card rates at 10%

Published

2 hours ago

January 10, 2026

Jace Porter

President Donald Trump on Friday called for a one-year cap on credit card interest rates at 10%, effective Jan. 20, without specifying details.

“Please be informed that we will no longer let the American Public be ‘ripped off’ by Credit Card Companies that are charging Interest Rates of 20 to 30%, and even more, which festered unimpeded during the Sleepy Joe Biden Administration. AFFORDABILITY!” he wrote on social media.

It’s not clear whether credit card companies will respond to his call, or what actions he might take to force any change.

The post comes as the Trump administration intensifies efforts to demonstrate to voters that the president is addressing concerns about costs and prices that have emerged as a central issue in the November midterm elections.

During the 2024 presidential campaign, Trump pledged to seek limits on the interest credit card companies can charge.

Hours before his message on Friday, Senator Bernie Sanders, a Vermont independent, said on X: “Trump promised to cap credit card interest rates at 10% and stop Wall Street from getting away with murder. Instead, he deregulated big banks charging up to 30% interest on credit cards.”

In a letter last year to Sanders and Senator Josh Hawley, a Missouri Republican, a group of banking trade groups painted a dire outcome for consumers if the government ever capped interest rates on credit cards at 10%, as the senators had proposed.

“Many consumers who currently rely on credit cards would be forced to turn elsewhere for short-term financing needs, including pawn shops, auto title lenders or worse — such as loan sharks, unregulated online lenders and the black market,” the group wrote.

The Bank Policy Institute said in a report last year that “while the proposed cap is a well-intentioned effort to reduce the high debt burden some households are facing, it would harm consumers’ access to card credit.” The group also said such a move could force card issuers to reduce cardholder benefits, including lucrative rewards tied to purchases.

Responding to Trump’s post on Friday, Hawley said on X: “Fantastic idea. Can’t wait to vote for this.”

Source link

Business

Asian households still save as much as half their wealth in cash. Fintech platforms like Syfe want to change that

Published

4 hours ago

January 9, 2026

Jace Porter

Growing up in India, Dhruv Arora’s mother gave him one key piece of financial advice: Put his money in the bank.

But Arora, now the founder of Singapore-based fintech platform Syfe, quickly realized that following his mother’s advice meant his money “did absolutely nothing.”

“We have quite a heavy culture of saving,” Arora says, citing Asia’s often unstable economic and policy history. But inflation and low interest rates end up eroding the value of household savings. “Over time, the $100 you put in the bank doesn’t become $101, but effectively $98” due to the effects of inflation.

Asian households sometimes keep as much as 50% of their net worth in cash, rather than in investments or assets. In contrast, in developed markets like the U.S. and Europe, that figure is closer to 15%.

But that conservative attitude in Asia is starting to change. Asians are getting wealthier, pushing them to explore different investment options. Strong stock market performance is also driving a new wave of retail investors across the Asia-Pacific.

“Asian households are slowly dipping their toes into stock markets,” HSBC economists wrote in a Jan. 9 report, though noted that “overall equity investment remains quite low.” The bank predicts that a steady shift from low-yield cash to higher-yield investments will mean “more money will continue to rotate into equity markets over the next few years,” reducing a reliance on foreign investors.

A slew of fintech apps have emerged in recent years to tap a growing interest in investing and wealth management among Asian users. These alternative finance platforms, such as Syfe, Stashaway and Endowus, often offer a range of investment options, ranging from cash management to managed portfolios and options trading. The challenge, Arora says, is how to “bridge the gap between holding money and growing wealth,” and “give more people the confidence to put their savings to work.”

Arora began his career as an investment banker for UBS in Hong Kong in 2008, soon after the Global Financial Crisis. Despite Asia’s relatively quick recovery, Arora noticed that the region’s professionals were building wealth yet didn’t know how to manage it. “These were smart people like doctors, lawyers and consultants, who were doing well professionally, but just did not know what to do with their money,” he says.

He launched Syfe in 2019, just a few months before another global crisis: The COVID-19 pandemic. Yet the pandemic ended up being an opportunity for fintech platforms like Syfe. “It acted as a catalyst for a shift in investor behavior,” Arora explained, as people suddenly had the time to engage with financial markets.

In the U.S., for example, people stuck at home began to get involved in stock trading through platforms like Robinhood. Fueled by social media, these retail investors began to heavily trade in so-called meme stocks like Gamestop and AMC.

Syfe has since expanded from its home market of Singapore to new Asia-Pacific economies like Australia and Hong Kong. The platform continues to grow both its userbase and company revenue, and the company claimed it reached profitability in Q4 2025. It’s now a “self-sustaining organization,” Arora says.

Syfe closed an $80 million Series C funding round last year, and is backed by major investors like NYC-based Valar Ventures and UK-based investment firm Unbound.

The platform’s users generated $2 billion worth of returns while saving $80 million in fees last year, according to the company.

Currently, Arora wants to deepen Syfe’s presence in its existing markets. Last year, the platform began to roll out bespoke offerings for its users, like private credit for accredited investors looking to diversify their portfolios on Syfe. Syfe will launch options trading in 2026.

Arora notes that many of Syfe’s users, over time, have grown more comfortable with taking larger investment risks, moving from putting their money in Syfe-managed portfolios, to more actively trading on brokerages and income portfolios.

Yet he eventually wants to bring Syfe to new markets in North Asia and the Middle East, which boast sizable populations of what Arora terms the “mass affluent,” a population with significant investable assets and higher-than-average incomes, though still not in the high-net-worth category.

“This demographic has historically been ‘stuck in the middle’: too large for basic retail banking, yet often underserved by traditional private banks,” he explains.

This story was originally featured on Fortune.com

Source link

Business

Lawmakers and victims criticize new limits on Grok’s AI image as ‘insulting’ and ‘not effective’

Published

4 hours ago

January 9, 2026

Jace Porter

Elon Musk’s xAI has restricted its AI chatbot Grok’s image generation capabilities to paying subscribers only, following widespread condemnation over its use to create non-consensual sexualized images of real women and children.

“Image generation and editing are currently limited to paying subscribers,” Grok announced via X on Friday. The restriction means the vast majority of users can no longer access the feature. Paying, verified subscribers with credit card details on file can still do so, but theoretically they can be identified more easily if the function is misused.

However, experts, regulators, and victims say that the new restrictions aren’t a solution to the now widespread problem.

“The argument that providing user details and payment methods will help identify perpetrators also isn’t convincing, given how easy it is to provide false info and use temporary payment methods,” Henry Ajder, a UK-based deepfakes expert, told Fortune. “The logic here is also reactive: it is supposed to help identify offenders after content has been generated, but it doesn’t represent any alignment or meaningful limitations to the model itself.”

The UK government has called the move “insulting” to victims, in remarks reported by the BBC. The UK’s prime minister’s spokesperson told reporters on Friday that the change “simply turns an AI feature that allows the creation of unlawful images into a premium service.

“It is time for X to grip this issue; if another media company had billboards in town centers showing unlawful images, it would act immediately to take them down or face public backlash,” they said.

A representative for X said they were “looking into” the new restrictions. xAI responded with the automated message: “Legacy Media Lies.”

Over the past week real women have been targeted at scale with users manipulating photos to remove clothing, place subjects in bikinis, or position them in sexually explicit scenarios without their consent. Some victims reported feeling violated and disturbed by the trend, with many saying their reports to X went unanswered and images remained live on the platform.

Researchers said the scale at which Grok was producing and sharing images was unprecedented as, unlike other AI bots, Grok essentially has a built-in distribution system in the X platform.

One researcher, whose analysis was published by Bloomberg, estimated that X has become the most prolific site for deepfakes over the last week. Genevieve Oh, a social media and deepfake researcher who conducted a 24-hour analysis of images the @Grok account posted to X, found that the chatbot was producing roughly 6,700 sexually suggestive or nudifying images per hour. By comparison, the five other leading websites for sexualized deepfakes averaged 79 new AI undressing images hourly during the same period. Oh’s research also found that sexualized content dominated Grok’s output, accounting for 85% of all images the chatbot generated.

Ashley St. Clair, a conservative commentator and mother of one of Musk’s children, was among those affected by the images. St. Clair told Fortune that users were turning images on her X profile into explicit AI-generated photos of her, including some she said depicted her as a minor. After speaking out against the images and raising concerns about deepfakes on minors, St Clair also said X took away her verified, paying subscribers status without notifying her or refunding her for the $8 per month fee.

“Restricting it to the paid-only user shows that they’re going to double down on this, placing an undue burden on the victims to report to law enforcement and law enforcement to use their resources to track these people down,” Ashley St Clair said of the recent restrictions. “It’s also a money grab.”

St Clair told Fortune that many of the accounts targeting her were already verified users: “It’s not effective at all,” she said. “This is just in anticipation of more law enforcement inquiries regarding Grok image generation.”

Regulatory pressure

The move to limit Grok’s capabilities comes amid mounting pressure from regulators worldwide. In the U.K., Prime Minister Keir Starmer has indicated he is open to banning the platform entirely, describing the content as “disgraceful” and “disgusting.” Regulators in India, Malaysia, and France have also launched investigations or probes.

The European Commission on Thursday ordered X to preserve all internal documents and data related to Grok, stepping up its investigation into the platform’s content moderation practices after describing the spread of nonconsensual sexually explicit deepfakes as “illegal,” “appalling,” and “disgusting.”

Experts say the new restrictions may not satisfy regulators’ concerns: “This approach is a blunt instrument that doesn’t address the root of the problem with Grok’s alignment and likely won’t cut it with regulators,” Ajder said. “Limiting functionality to paying users will not stop the generation of this content; a month’s subscription is not a robust solution.”

In the U.S., the situation is also likely to test existing laws, like Section 230 of the Communications Decency Act, which shields online providers from liability for content created by users. U.S. Senators Ron Wyden, Edward J. Markey, and Ben Ray Luján have issued a statement urging Apple and Google to “immediately remove the X and Grok apps from their app stores” following Grok’s alleged use for generating “nonconsensual sexualized images of women and children at scale.” The lawmakers called the images “disturbing and likely illegal,” and said the apps should remain unavailable until Musk addresses the concerns.

The Council on American-Islamic Relations (CAIR) has also called for Grok to be blocked from generating “sexually explicit images of children and women, including prominent Muslim women.”

Riana Pfefferkorn of Stanford’s Institute for Human-Centered Artificial Intelligence previously told Fortune that liability surrounding AI-generated images is murky. “We have this situation where for the first time, it is the platform itself that is at scale generating non-consensual pornography of adults and minors alike,” she said. “From a liability perspective as well as a PR perspective, the CSAM laws pose the biggest potential liability risk here.”

Musk has previously stated that “anyone using Grok to make illegal content will suffer the same consequences as if they upload illegal content.” However, it remains unclear how accounts will be held accountable.

Source link