Anthropic’s sudden move to suspend access to its newest AI models following a U.S. government directive has raised fresh questions across the global technology industry. In India, the decision has reignited a long-running debate over whether one of the world’s largest AI markets can afford to rely on technologies built and controlled elsewhere.

The announcement came late Friday, when Anthropic said it had received the U.S. government directive requiring it to suspend access to its recently launched Fable 5 and Mythos 5 models for all foreign nationals, including its own foreign national employees. The move came shortly after the company announced a partnership with Indian IT services giant Tata Consultancy Services to expand enterprise AI adoption in India, underlining how closely the country’s AI ambitions have become tied to technologies developed and governed in the U.S.

While the broader implications remain unclear, some reports said the initial security concerns were first reported to the government by Amazon CEO Andy Jassy. And The Information said the White House is unlikely to extend similar restrictions to other AI companies and is privately blaming Anthropic’s handling of alleged jailbreak vulnerabilities. Anthropic has disputed the government’s characterization and argued the action should not have been taken.

Regardless, the development has triggered debate among Indian founders, investors, and policy experts over whether the country should accelerate efforts to build domestic AI capabilities, deepen investment in open-source alternatives, or continue relying on a handful of U.S. frontier model providers. For some, the episode is a wake-up call on technological dependence. For others, it is a reminder that access to increasingly critical AI systems can be shaped by geopolitical decisions beyond India’s control.

India has become one of the most important markets for frontier AI companies. Anthropic and OpenAI have both described the South Asian nation as their second-largest market after the U.S., reflecting its growing importance in the global AI race. The companies have already set up their offices in India, expanded local hiring, partnerships, and enterprise initiatives in recent months, betting on India’s vast base of developers, startups, and businesses to accelerate adoption of their latest technologies.

For many in India’s technology sector, Anthropic’s Friday announcement was about more than just one AI company. It reopened questions about the country’s long-term AI strategy and whether India could afford to remain dependent on a small number of foreign frontier AI providers.

“It completely changes things,” said Aakrit Vaish, founder of Indian AI venture platform Activate, referring to Anthropic’s decision. “I think this materially changes the way all of us should be thinking about sovereign AI in India.”

Vaish told TechCrunch that he woke up on Saturday morning “shocked and confused” by the announcement and said it strengthened the case for developing domestic AI capabilities. He expects startups to increasingly turn to open-source models and plans to encourage companies in his portfolio to reduce their dependence on a small number of frontier AI providers.

For some founders, the bigger concern was what restrictions on frontier AI access could mean for competitiveness. Vijay Rayapati, co-founder and CEO of Atomicwork, told TechCrunch that the episode highlighted the risks facing startups whose teams span multiple countries if access to advanced AI systems increasingly becomes subject to geopolitical restrictions.

Atomicwork has around 25 employees in the U.S., though much of its product engineering team is based in Bengaluru, India.

“If your AI team is not made up entirely of U.S. citizens, you are at a competitive disadvantage,” Rayapati said, arguing that unequal access to frontier AI models could give some companies a significant edge over rivals.

The concern comes as parts of India’s tech sector are already grappling with questions about how AI could reshape the economics of global talent. This week, U.S. real estate technology company Opendoor shut its India office less than two years after expanding in the country, with CEO Kaz Nejatian citing a push to bring operational work closer to customers in the U.S. and a shift toward smaller AI-native teams.

While Opendoor did not specify how much of the decision was driven by AI-related efficiencies, the move added to a broader debate about how advances in AI could affect the future of global technology work and what that might mean for India’s position as an engineering talent hub.

Beyond Anthropic

In addition to startups and AI builders, the Anthropic episode also prompted a broader debate among India’s technology leaders about dependence on foreign AI infrastructure.

Sridhar Vembu, founder of Indian SaaS company Zoho, said the move showed that “technology is the ultimate weapon” and urged Indian organizations to increasingly embrace smaller and open-source models.

“What can our government do right now? Ensure that orgs in India embrace smaller models, both Indian and Chinese open source ones,” Vembu wrote on X.

Investor and former Infosys executive Mohandas Pai responded to Vembu on X, arguing that the development highlighted the need for a far more ambitious national AI strategy and calling on the government to substantially increase investments in AI, computing infrastructure, and deep technology.

“We are way behind and need a national mission to get going quickly,” Pai wrote, urging the government to create an annual ₹500 billion (about $5 billion) fund for AI and deep tech, alongside a ₹2 trillion (around $21 billion) credit guarantee program to support cloud infrastructure, hardware, and semiconductor development.

Pai’s proposal would dwarf India’s existing AI efforts. In 2024, New Delhi approved the IndiaAI Mission with an outlay of ₹103.72 billion (about $1.2 billion) over five years, aimed at expanding compute infrastructure, supporting startups, and developing indigenous AI capabilities.

Despite growing interest in AI and New Delhi’s push to develop domestic capabilities, India remains a relatively small player in frontier model development. Only a handful of startups are pursuing foundational AI models, including Sarvam, which released open-source models earlier this year. However, another high-profile AI startup, Krutrim, pivoted toward cloud and AI infrastructure services after initially positioning itself around foundational model development.

Much of India’s AI ecosystem has instead concentrated on applications and specialized models built on top of existing foundation models. Recent examples include Avataar AI, which launched a video-generation model earlier this week aimed at providing a lower-cost alternative to offerings from rivals including Google’s Veo, Kling, Luma, and Runway.

Not everyone agrees that the primary challenge is a lack of capital. Responding to Pai’s comments, Lightspeed partner Hemant Mohapatra argued that the biggest constraints to building globally competitive AI companies are talent, access to computing resources, and execution, rather than simply the size of investment commitments.

Mohapatra estimated that training a frontier AI model could cost anywhere from hundreds of millions to several billion dollars, depending on the approach, but said successful AI companies have historically scaled their capital requirements over time as adoption grew.

Yet for some policy observers, the implications extend well beyond AI startups or model providers.

Prasanto Roy, a New Delhi-based technology policy expert who advises multinational companies, said the episode would likely reinforce concerns within the Indian government about strategic autonomy, comparing it to the lesson many countries drew from Russia’s loss of access to SWIFT and other parts of the global financial system following its invasion of Ukraine.

He told TechCrunch that the move was likely to provoke a significant nationalist backlash in India and described it as a poorly considered decision by Washington, with consequences extending far beyond Anthropic itself.

“Even if this is corrected or reversed, the Anthropic episode shows there’s no such thing as a geopolitically neutral foreign LLM,” Roy said. “American AI models are bound to American geopolitics.”

Artificial intelligence has been one of the most influential forces shaping recruitment in a global talent war. The volume of data companies can now access, the speed at which candidate pools can be filtered, and the complexity of searches that can be executed in minutes; these are all genuine advances. Yet amid the enthusiasm surrounding automation, Michael Ronis, founder of Janbrook Partners, believes many companies are asking the wrong question.

“AI opens a lot of doors in the sense of access to information,” Ronis says. “You can now research things and approach searches in a much more complex way than you could in the past. The real game changer is the access to information and the ability to break down information.”

Ronis argues that the debate today is largely focused on whether AI can replace recruiters. He rejects that conversation entirely. Instead, the more vital consideration, in his view, lies in a single question: at what point does a human step in?

Recent surveys show that 88% of employers now incorporate AI to accelerate talent acquisition and candidate screening. The appeal, he notes, is easy to understand as the volume problem alone makes some degree of automation non-negotiable. With companies receiving over a million applications in a year, Ronis notes that AI, at that scale, becomes a necessity, as it would be nearly impossible to manage that quantity manually.

Ronis has experienced that reality firsthand. “We ran an ad for a remote recruiter position and got a thousand resumes in a matter of hours,” he says. “At that point, give me the AI.”

Yet he believes the industry has conflated efficiency with effectiveness, and the financial consequences of that, Ronis adds, are catching up. Replacing an employee can cost anywhere from 50% to 200% of their annual salary, once recruitment fees, onboarding, training, and lost productivity are factored in. At the same time, Ronis observed that employee turnover continues to challenge employers across industries, particularly during the first year of employment.

According to Ronis, recruitment should not be measured solely by how quickly a role is filled. The more meaningful metric is how successfully that hire performs and remains with the organization over time.

Many companies focus on reducing hiring costs through automation, he argues, while paying less attention to the financial consequences of poor retention. “If you automate things to the extent that you overlook cultural fit, you wind up with a situation where you buy it cheap and buy it twice,” Ronis says.

His concern is not that AI lacks analytical capability. Rather, it lacks the ability to evaluate the nuanced human dynamics that often determine whether a candidate succeeds. “At the end of the day, recruiting is relationships,” he explains. “The numbers can get you so far. They can’t replicate rapport. They can’t give you some of the things that hiring decisions are ultimately made of.”

Those dynamics become increasingly important as candidates move deeper into the recruitment process. AI may identify qualifications, rank applicants, and surface relevant profiles, but Ronis points out that it cannot reliably determine a candidate’s level of commitment, career motivations, interpersonal style, or alignment with a company’s working environment.

“You can wind up deeply in the process with someone who might not be all that interested, or who has wildly different salary expectations,” Ronis says. “AI can only do so much. At a certain point, you have to take over.”

Trust also plays a role. As organizations increase their reliance on automated systems, Ronis believes many job seekers have grown skeptical about whether applications are receiving meaningful consideration. “The candidates don’t trust the process,” he says. “They don’t believe their resume is being seen a lot of the time.”

In his view, this growing perception can create challenges for employers seeking to build strong candidate relationships. This erosion of trust can have consequences for employer reputation and candidate engagement, particularly in competitive hiring markets where top talent often has multiple options.

Ronis views cultural fit as another area where human judgment remains indispensable. Every organization has unique interpersonal dynamics and workplace expectations that cannot be fully captured through algorithms or keyword matching. “There are dynamics that underlie every office environment. You have to find the balance where it fits. That’s what the hiring process is really about in the first place. Finding the person who fits the role the best,” he explains.

From his perspective, the strongest recruitment strategies are built around understanding how both can complement one another. AI can accelerate research, uncover patterns, and help recruiters navigate overwhelming volumes of information. Meanwhile, human professionals can contribute judgment, intuition, relationship-building skills, and the ability to assess qualities that may not appear on a resume.

Ronis remarks, “The difference is the human being behind it. Just because the information exists doesn’t mean it automatically creates the right outcome.”

Recruitment has always been about finding the right person, not simply processing applications faster. Technology may improve the search, but Ronis emphasizes that discernment remains the element that turns a candidate into a successful long-term hire.

Hidden among the architectural landmarks of General Motors’ sprawling Warren Tech Center outside Detroit is a new cornerstone of the automaker’s $900 million bet on its electric future.

The nondescript 500,000-square-foot pair of off-white boxes, which house GM’s new Battery Cell Development Center, might not look like much. But locked inside is the key to GM’s plan to slash the cost of its EVs by nearly 10%. 

At a time when some car companies are pulling back on EVs, GM’s new Battery Cell Development Center is part of a reboot. And it’s one that GM told TechCrunch will allow it to bring a new slate of lower-cost batteries to market a year faster than planned.

GM hasn’t been immune to the malaise in the U.S. EV market. Last year, the automaker took a $1.6 billion charge as it reconfigured its EV production capacity, laying off thousands of workers in the process. It has also reportedly shelved, if temporarily, a refresh of its full-size EV trucks and SUVs.

To get its EV strategy back on track, Kurt Kelty, vice president of battery and sustainability at GM, is pinning the company’s success on a new battery chemistry known as LMR. Kelty, who previously led battery technology at Tesla, has made it his signature product in the two years he’s been with the company.

“That is really going to be our bread and butter,” Kelty told TechCrunch. “That is going to be our main product line.”

Battery reboot

GM’s halting rollout of EVs has mirrored the wider battery industry in the U.S., which over the last couple of decades has developed in fits and starts. Early startups haven’t lived up to their promise, and more recently, intense competition from Chinese companies has pushed automakers and battery manufacturers to rethink the plans they made five years ago.

At GM, that pressure led to the shortened life of Ultium, the branded battery platform that underpins its current EVs. Like much of the industry, the automaker had bet heavily on a pricey yet powerful battery chemistry known as NMC (nickel-manganese-cobalt). Rising materials costs and China’s dominance of key critical minerals have kept EV prices higher than expected. NMC won’t disappear, but at GM, it’ll be restricted to GM’s high-end vehicles.

In its place, GM has been developing LMR (lithium-manganese-rich), which it says is almost as energy dense as NMC but at a cost that’s comparable to cheaper chemistries like LFP (lithium-iron-phosphate) that power low-end models like the Chevrolet Bolt.

When GM introduced LMR last year, it said that, in a truck like the Chevrolet Silverado EV, the new chemistry should preserve most of the vehicle’s more than 400-mile range while slashing costs by at least $6,000. For a mid-range model, that would bring it within spitting distance of the gas version.

Discovering a new battery chemistry is one thing. Manufacturing gigawatt-hours’ worth of it is another, especially at the pace the EV industry is moving. Facing pressure from automotive giants like BYD and battery titans like CATL, GM says it wants to get LMR vehicles on the road by 2028. GM needs the new Battery Cell Development Center to deliver if it wants to hit that deadline.

The new building serves as the keystone of GM’s battery strategy. The company opened its Wallace Battery Cell Innovation Center and its first gigafactory in 2022. What was missing was a way to connect the breakthroughs that emerged from Wallace to the factory floors in Tennessee and Ohio.

The BCDC, as insiders call the facility, is something like a pilot line, but bigger. When fully operational, it will be capable of producing about 2,500 cells per day, or about half a gigawatt-hour per year. It will take batteries developed in small batches — about 30 to 50 per day — at the Wallace Battery Cell Research Center next door and determine if they’re ready for production. 

Mastering the battery recipe

Many recipes for new batteries fail to deliver when they’re spun up to commercial scale, and companies don’t have years to work out the kinks. If a new chemistry can’t hit 85% yield within 18 months on a production line, it shouldn’t be considered commercially viable, according to a McKinsey report.

The challenges are similar to using a recipe intended for a family of four and scaling it up to a wedding reception with 400 guests. It’s not just the sheer throughput of the factory, either. Batteries that emerge from the research center are small coin cells, but the cells in an EV pack look more like a small cutting board.

“Once you learn how to make the recipe in Wallace, then you’ve got to figure out, well, how do you make this in high volume?” Kelty said. “You really learn a lot going from that coin cell to the large format because it doesn’t transfer perfectly.”

The BCDC is intended to make that step less painful. 

A test run at the facility costs about $200,000, which is far less than at the full-size Ultium plant. When the BCDC team is confident it has the process nailed down, the transition to full production should be easier, Kelty said. “The equipment is almost the same between them, and so it shouldn’t be as hard of a handoff.”

The BCDC is one or two orders of magnitude smaller than the 2.8 million-square-foot Ultium battery factory in Tennessee. The Ultium plant makes about 300,000 cells per year, or 45 gigawatt-hours’ worth. The BCDC has fewer production lines, makes about a hundredth the number of cells, and its mixing tanks, where battery materials are blended, hold 40 liters instead of 2,000. Though smaller, the BCDC is still an order of magnitude larger than the Wallace Center next door. 

“The BCDC is intended to bridge the gap,” Mo Gallegos, head of BCDC at GM, told TechCrunch. 

Turning to AI models

To cut costs further, GM has been working to simulate as many processes as possible using a variety of AI models. The company has invested heavily in computing power, and while no one would put a number on it, I’m told it’s “national lab-scale.”

The automaker has developed physics-based models to simulate how changes to a chemistry or production process will affect the performance of a battery cell. 

“On LMR, we’ve logged over 150 million CPU hours,” Radu Theyyunni, director of global virtual electrification and powertrain at GM, told TechCrunch. “Most engine programs do not use that many core hours.” 

There’s also a digital twin of the entire BCDC, including equipment control boards, wiring, and even the blades in the mixing tanks. Before I set foot in the BCDC, the team had me don a VR headset and walked me through the digital twin, where I was able to follow the production line from start to finish.

As the BCDC has taken shape, the digital twin has been used for a range of tasks. In one instance, the team used it to determine if the plans left enough clearance around equipment for regular operations and repairs. In another, they simulated the equipments’ control systems to ensure everything would behave as intended. 

“Does the equipment run how it’s supposed to? Does it run safely? Is it doing all the things we think this control system is going to do? That shortens our debug and ramp up time,” Gallegos said. Altogether, GM says the simulations have saved it millions of dollars.

GM needs all the speed it can get. 

While the EV market in the U.S. has softened recently, globally, it grew 20% last year. The looming specter of high oil prices coupled with declining battery costs suggests the transition away from fossil fuels will happen eventually, if not sooner. 

If LMR is ready in time, it could help GM offer cost-competitive EVs with enough range to placate anxious Americans. But first LMR needs to pass through the BCDC. Gallegos expects the first batches to roll off the line later this year.

In the coming decade, battery development will be as important to automakers as engine development was over the last century. GM’s EV future hinges on its ability to shepherd new chemistries from R&D through to production. 

Kelty is fond of saying that GM is developing “the right battery for the right application,” perhaps echoing an old company slogan, “a car for every purse and purpose.” 

LMR might be the BCDC’s first test, but it’s unlikely to be its last.

SpaceX is warning investors that it may dole out “significant equity” in “future transactions” following its upcoming IPO — language that was added amid constant industry gossip that CEO Elon Musk ultimately plans to merge his space-and-AI company with Tesla.

The new language was tucked deep in the risk factors section of SpaceX’s first official amendment to its IPO filing, which was made public last month. The company tacked this sentence onto the end of the first paragraph of a risk about how mergers and acquisitions may go sideways:

We may issue a significant amount of equity in connection with future transactions.

SpaceX has been busy with M&A, acquiring Musk’s AI company xAI last year, and recently entering into a deal with Cursor that includes an option to buy the startup for $60 billion in stock following the IPO. It’s certainly possible that SpaceX will have other targets in mind after it raises a reported $75 billion when it lists on the Nasdaq exchange (minus $20 billion committed to paying down former xAI and X debt). But this warning seems tailored to prepare investors for the possibility of a major dilution event — like a future combination with Tesla.

Musk has mused about combining his companies for many years, and the SpaceX IPO is only heating up the chatter that he will finally bring his two biggest entities together. A merger of this size would face a number of legal and potential regulatory challenges, and would likely have to pass a shareholder vote at Tesla. But, as the IPO filing showed, Musk has supreme voting power at SpaceX; the only person who could vote down a merger on that side of the term sheet would be Musk.

Musk’s voting power at SpaceX would not be at risk during a major dilution event. SpaceX has three primary classes of shares heading into this IPO. They all have the same basic economic rights, but different voting rights.

The Class A shares are what will be sold to the public and come with one vote per share. The Class B shares are owned exclusively by Musk and have 10 votes per share. SpaceX also has Class C common stock, which comes with no voting rights at all. While those Class C shares are currently used for executive compensation, Musk could use those shares to buy other companies without diluting his power. (SpaceX also has Class D shares set aside, which have reduced economic rights; the company has not yet decided whether those shares will have any voting power.)

Some of the most consequential companies in tech history didn’t launch with a splashy fundraising announcement. They started with a pitch. Dropbox demoed to a room of skeptics. Cloudflare took the stage before most people understood what edge networking meant. Discord was a scrappy game developer called Hammer & Chisel. Mint, Trello, Forethought, N26 — all of them passed through the same crucible: TechCrunch Startup Battlefield.

That’s not a coincidence. Battlefield isn’t just a competition. It’s a launchpad, and the numbers back it up. More than 1,700 companies have competed on the Battlefield stage. Together, they’ve raised $32 billion in total funding and generated over 250 exits — including acquisitions by Microsoft, Google, Yahoo, Salesforce, Twitter, Uber, and Amazon. The Startup Battlefield network runs so deep that alumni have even acquired each other: Dropbox acquired fellow Startup Battlefield alum DocSend in 2021. For thousands of founders, it’s become a defining milestone — not just a pitch competition, but the moment the world started paying attention.

And you actually still have a chance to join that illustrious alumni community this year. Due to intense demand, we’ve pushed the Startup Battlefield 2026 application deadline to June 8, and you can get your application in right here.

In the meantime, we wanted to show you what happens after the confetti falls. We checked in with some of our recent alumni, many of whom have sat down with us on Build Mode: The Founder Survival Guide, TechCrunch’s podcast for founders at every stage. Here’s what they’ve been building, in their own words.

About Build Mode

Each season goes deep on a different chapter of startup life. Season 1 covered go-to-market. Season 2 — out now — is all about building your team. And mark your calendars: Season 3 drops in June, tackling the most requested topic we’ve ever gotten: fundraising.

Subscribe now so you don’t miss it.

The champions and runners-up

From military logistics to Startup Battlefield 2025 champion

Kevin Damoa, founder of Glīd — 2025 winner

Kevin Damoa didn’t come from Sand Hill Road. He came from military logistics — a background that turned out to be ideal training for building under pressure, with constrained resources and real stakes. Damoa’s path to the Startup Battlefield 2025 championship is the kind of origin story that makes you reconsider where the next generation of great founders is actually coming from.

→ Listen to Kevin’s Build Mode episode

From the Startup Battlefield stage to the International Space Station

Capella Kerst, founder and CEO of geCKo Materials — 2024 runner-up

Capella Kerst didn’t set out to reinvent adhesion. She set out to solve a problem that has stumped engineers for decades: How do you make things stick — reliably, repeatedly, and without residue — in the most extreme environments imaginable? geCKo Materials, spun out of Stanford, has developed gecko-inspired adhesive technology with applications ranging from manufacturing floors to, quite literally, the International Space Station.

Kerst’s Startup Battlefield moment was a signal to the market that the science was ready for the world. What’s happened since is proof that runner-up isn’t a consolation prize — it’s a credential. Hear how she got there:

→ Listen to Capella’s Build Mode episode

How Forethought AI found product-market fit — before it was obvious

Deon Nicholas, co-founder of Forethought AI — 2018 winner (acquired by Zendesk) 

Few Startup Battlefield stories have a more complete arc than Forethought AI. Deon Nicholas took the stage with a conviction that AI could fundamentally transform customer support — before that was an obvious bet. Before the term sheets and the headlines, there was a pitch and a thesis. Forethought was recently acquired by Zendesk — the latest example of what the Startup Battlefield stage can set in motion. His Build Mode episode is essential listening, and a perfect primer for Season 3’s deep dive on fundraising.

→ Listen to Deon’s Build Mode episode

Top 20 finalist stories

The danger of fundraising before finding product-market fit 

David Park, founder of Narada

Raising before product-market fit doesn’t speed things up — it speeds up your mistakes. Park doesn’t sugarcoat the lessons.

→ Listen to David’s Build Mode episode

Using AI to hire for compatibility, not just skill

Sarah Lucena, founder and CEO of Mappa

Skills get people in the door. Compatibility determines whether they stay. Lucena is using AI to fix the part of hiring nobody talks about.

→ Listen to Sarah’s Build Mode episode

These founders competed on the Startup Battlefield and sat down with us on Build Mode to tell their story. All worth a listen.

Anna Sun of Nowadays and Hala Jalwan and Alessio Tresanti of Rivio — On what happens when a startup becomes a family business, and the community that forms around Startup Battlefield. → Listen

Kyle Rudolph and Jon Walburg, co-founders of Alltroo — On why your network is your first go-to-market strategy. → Listen

Jas Schembri-Stothart of Luna and Andre Peart of Untapped Solutions — On reaching the markets everyone else ignores and building for underserved communities without the typical growth playbook. → Listen

The milestone is real

Every generation of Startup Battlefield alumni adds a new chapter to the same story. But behind every one of those data points is a founder who made a bet on themselves — publicly, in front of people who were paying attention. The stage matters. The community lasts. The milestone is real.

And remember: Applications for Startup Battlefield 2026 are still open. If you’re building something that deserves a stage, this is yours.

→ Apply before the June 8 deadline

Know a founder who’s ready for the spotlight? Investors, operators, and fellow founders can nominate companies directly.

→ Nominate a founder

Not ready to apply yet? Build Mode is where we meet you. Season 2 is live now. Season 3 — all about fundraising — drops this summer.

→ Subscribe to Build Mode