The Walls Within: Why Organizations Cling to Data Silos in the Age of AI: By Erica Andersen
Yet, despite this allure, organizations are often hesitant to embrace the full power of AI across the entire enterprise. Instead, we see a persistent trend: the deliberate creation and maintenance of data silos, where information remains walled off, and AI's access is carefully restricted. This isn't necessarily a sign of technological backwardness or a lack of vision. Rather, it's a complex tapestry woven with threads of business strategy, legal compliance, technical limitations, and ingrained organizational culture. This article delves into the multifaceted reasons behind this phenomenon, exploring why organizations are choosing to keep their AI contained within the familiar confines of their data silos.
The Security Fortress: Protecting Data in a Vulnerable World
At the heart of this reluctance lies a deep-seated concern for data security and privacy. Organizations are acutely aware of the potential for catastrophic data breaches, and the implications are severe.
Protecting Sensitive Information: The risk of exposing sensitive information like Personally Identifiable Information (PII), financial records, trade secrets, and intellectual property is a constant threat. Restricting access is a fundamental strategy to minimize the "attack surface" and reduce the likelihood of a breach. This includes not only protecting against malicious actors but also accidental disclosures, which can have significant legal and reputational consequences.
The risk of exposing sensitive information like Personally Identifiable Information (PII), financial records, trade secrets, and intellectual property is a constant threat. Restricting access is a fundamental strategy to minimize the "attack surface" and reduce the likelihood of a breach. This includes not only protecting against malicious actors but also accidental disclosures, which can have significant legal and reputational consequences. Compliance is King: Navigating the Regulatory Minefield: Regulations like GDPR (General Data Protection Regulation), CCPA (California Consumer Privacy Act), HIPAA (Health Insurance Portability and Accountability Act), LGPD (Lei Geral de Proteção de Dados - Brazil), and industry-specific mandates demand robust data privacy and security measures. Maintaining data silos is often seen as a practical way to simplify compliance by limiting the scope of data that needs to be protected.
Regulations like GDPR (General Data Protection Regulation), CCPA (California Consumer Privacy Act), HIPAA (Health Insurance Portability and Accountability Act), LGPD (Lei Geral de Proteção de Dados - Brazil), and industry-specific mandates demand robust data privacy and security measures. Maintaining data silos is often seen as a practical way to simplify compliance by limiting the scope of data that needs to be protected. Unauthorized Access: A Primary Threat: Data silos create physical and logical barriers, making it significantly harder for unauthorized individuals or external actors to access and potentially misuse sensitive data. This includes implementing robust access controls, multi-factor authentication, and regular security audits.
Data silos create physical and logical barriers, making it significantly harder for unauthorized individuals or external actors to access and potentially misuse sensitive data. This includes implementing robust access controls, multi-factor authentication, and regular security audits. Ethical Usage: Maintaining Control and Addressing Bias: Organizations want to ensure their data is used ethically and in accordance with their policies. Restricting access to AI models is a key mechanism for enforcing this control. This includes: Bias Detection and Mitigation: AI models can perpetuate biases present in the training data. Silos allow for careful curation of data and the application of bias detection and mitigation techniques. Explainability and Transparency: Organizations must be able to explain how their AI models make decisions. Silos can facilitate the development of explainable AI (XAI) by limiting the complexity of the data and the scope of the models. Accountability and Responsibility: Clearly defined roles and responsibilities are crucial for AI governance. Silos can help establish clear lines of accountability for data usage and model performance.
Organizations want to ensure their data is used ethically and in accordance with their policies. Restricting access to AI models is a key mechanism for enforcing this control. This includes:
The Competitive Edge: Data as a Strategic Weapon
Beyond security, the desire to protect competitive advantage and intellectual property is another driving force behind data silo maintenance.
Proprietary Data: The Secret Sauce: Data can be a valuable asset. Organizations may want to keep their unique data private to maintain a competitive edge. AI models trained on distinctive datasets can be a significant differentiator. This requires careful consideration of data licensing, access controls, and the potential for reverse engineering of AI models.
Data can be a valuable asset. Organizations may want to keep their unique data private to maintain a competitive edge. AI models trained on distinctive datasets can be a significant differentiator. This requires careful consideration of data licensing, access controls, and the potential for reverse engineering of AI models. Trade Secrets: Guarding the Jewels: The data used to train AI models can reveal valuable insights and trade secrets, offering competitors a roadmap to replicate innovations. Restricting access helps prevent reverse-engineering and exploitation. This includes implementing strict non-disclosure agreements (NDAs) and protecting the intellectual property rights associated with the AI models and the underlying data.
The data used to train AI models can reveal valuable insights and trade secrets, offering competitors a roadmap to replicate innovations. Restricting access helps prevent reverse-engineering and exploitation. This includes implementing strict non-disclosure agreements (NDAs) and protecting the intellectual property rights associated with the AI models and the underlying data. Data Leakage: Preventing Spills: Data silos act as barriers against data leakage, preventing valuable proprietary information from falling into the hands of competitors or external parties. This includes implementing robust data loss prevention (DLP) measures and monitoring for suspicious data activity.
The Governance Imperative: Maintaining Control and Quality
Organizations also prioritize control and governance over their data, recognizing the crucial role these play in the success of AI initiatives.
Data Quality: A Foundation for Success: Organizations want to maintain control over the quality of the data used for AI training. Silos allow for better data governance and quality control within each department or function. This includes implementing data validation rules, data cleansing processes, and data governance frameworks.
Organizations want to maintain control over the quality of the data used for AI training. Silos allow for better data governance and quality control within each department or function. This includes implementing data validation rules, data cleansing processes, and data governance frameworks. Accuracy and Reliability: The Pillars of Trust: Data accuracy and reliability are critical for AI model performance. Silos can help ensure that the data used for training is accurate and reliable, reducing the risk of biased or inaccurate results. This includes implementing data quality metrics, data lineage tracking, and data auditing processes.
Data accuracy and reliability are critical for AI model performance. Silos can help ensure that the data used for training is accurate and reliable, reducing the risk of biased or inaccurate results. This includes implementing data quality metrics, data lineage tracking, and data auditing processes. Responsible AI: Managing the Lifecycle: Restricting access to data allows organizations to better manage the development, deployment, and monitoring of AI models. This helps ensure that models are used responsibly and ethically. This includes: Model Monitoring: Continuously monitoring AI model performance and identifying potential issues, such as drift or bias. Model Versioning: Tracking different versions of AI models and the associated data used for training. Model Auditing: Regularly auditing AI models to ensure compliance with regulations and ethical guidelines.
Restricting access to data allows organizations to better manage the development, deployment, and monitoring of AI models. This helps ensure that models are used responsibly and ethically. This includes:
The Technical Hurdles: Navigating the Complexities
Beyond the strategic and legal aspects, technical and practical considerations also contribute to the prevalence of data silos.
Integration Challenges: A Complex Undertaking: Integrating data from multiple sources can be incredibly complex and time-consuming. Organizations may lack the necessary infrastructure, skills, or resources to effectively integrate data across silos. This includes challenges related to data format compatibility, data semantics, and data governance.
Integrating data from multiple sources can be incredibly complex and time-consuming. Organizations may lack the necessary infrastructure, skills, or resources to effectively integrate data across silos. This includes challenges related to data format compatibility, data semantics, and data governance. Data Standardization: A Formidable Task: Data from different sources may be in different formats or use different standards, making integration a challenging undertaking. This requires implementing data standardization processes, data transformation tools, and data governance frameworks.
Data from different sources may be in different formats or use different standards, making integration a challenging undertaking. This requires implementing data standardization processes, data transformation tools, and data governance frameworks. Scalability and Performance: Managing the Volume: Integrating and processing large volumes of data can strain infrastructure and impact performance. Silos can help manage data volume and improve performance. This requires implementing scalable data storage solutions, data processing frameworks, and data optimization techniques.
Integrating and processing large volumes of data can strain infrastructure and impact performance. Silos can help manage data volume and improve performance. This requires implementing scalable data storage solutions, data processing frameworks, and data optimization techniques. Legacy Systems: The Weight of History: Many organizations have legacy systems and infrastructure that are not designed for easy data sharing, adding another layer of complexity. This requires modernizing legacy systems, implementing data integration solutions, and gradually migrating data to more modern platforms.
The Human Factor: Navigating Organizational Dynamics
Finally, organizational culture and politics play a significant role in the decision to maintain data silos.
Departmental Autonomy: Protecting Territories: Departments or business units may want to maintain their autonomy and control over their data, viewing it as a valuable resource. This requires fostering a culture of collaboration, promoting data sharing best practices, and establishing clear data governance frameworks.
Departments or business units may want to maintain their autonomy and control over their data, viewing it as a valuable resource. This requires fostering a culture of collaboration, promoting data sharing best practices, and establishing clear data governance frameworks. Fear of Misuse: A Valid Concern: Some individuals or teams may be hesitant to share their data due to concerns about how it will be used or the potential for negative consequences. This requires establishing clear data usage policies, implementing data access controls, and providing training on responsible AI practices.
Some individuals or teams may be hesitant to share their data due to concerns about how it will be used or the potential for negative consequences. This requires establishing clear data usage policies, implementing data access controls, and providing training on responsible AI practices. Lack of Trust: A Barrier to Collaboration: There may be a lack of trust between different departments or teams, making them unwilling to share data. This requires building trust through open communication, transparency, and collaborative projects.
There may be a lack of trust between different departments or teams, making them unwilling to share data. This requires building trust through open communication, transparency, and collaborative projects. AI Anxiety: A Shift in Power: A department might fear that sharing data will lead to a loss of control or power, or that AI will replace human workers. This requires addressing these concerns through clear communication, providing training on AI technologies, and demonstrating the benefits of AI for both individuals and the organization as a whole. Highlighting how AI can augment human capabilities and improve job satisfaction is crucial.
In Summary: A Delicate Balance
The desire to maintain data silos in the context of AI adoption is a complex issue driven by a combination of factors, including data security, competitive advantage, regulatory compliance, technical challenges, and organizational culture. While data silos can offer benefits in terms of control and security, they can also hinder innovation and limit the potential of AI. Organizations must carefully weigh these competing considerations when developing their AI strategies, striving to find a balance that maximizes the benefits of AI while mitigating the risks. The future of AI adoption lies in finding innovative ways to navigate these complexities, fostering collaboration while safeguarding the valuable assets that organizations hold within their walls. This includes exploring strategies such as:
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Reuters
26 minutes ago
- Reuters
US bank M&A hopes revive under Trump regulators
NEW YORK, July 14 (Reuters) - Takeover speculation in Northern Trust (NTRS.O), opens new tab has revived industry hopes of deals among large U.S. and regional banks, propelling exploratory conversations that could lead to consolidation, according to financial executives and analysts. Talk of potential mergers and acquisitions among Wall Street banks and large regional lenders has increased in recent weeks in a major shift under the Trump administration after regulators under the Biden administration opposed or blocked big deals, according to three senior financial executives who declined to name specific talks or be identified, citing confidential discussions. On Thursday, the Federal Reserve proposed changes to how it evaluates large banks, making it easier for firms to maintain a "well managed" rating by requiring deficiencies across multiple categories before being downgraded. The move could be a boon to bigger bank dealmaking, as firms not considered "well managed" are barred from any acquisitions. "What we've seen from a regulatory standpoint is a lot more clarity and ... a return to a more permissive environment," particularly for mergers, said James Stevens, a law partner who advises financial institutions at Troutman Pepper Locke. Regulators' moves to streamline deal approvals "certainly opened the doors more towards those bigger banks talking about getting together," he said. The sources said that bank executives in recent weeks have become newly emboldened to consider ambitious plans to buy business units, or even entire companies. That increased interest came after BNY (BK.N), opens new tab approached Northern Trust (NTRS.O), opens new tab to express interest in a merger, the Wall Street Journal reported last month, although the target has said it wants to remain independent. Meanwhile regulators approved Capital One's (COF.N), opens new tab $35.3 billion purchase of Discover Financial Services in April. BNY will report earnings on Tuesday alongside JPMorgan, Wells Fargo and Citigroup. The companies will likely be quizzed about their appetite for M&A during analyst calls. BNY and Northern Trust declined to comment. Dealmakers expect bank M&A activity to climb in the second half of the year. Activity has been broadly flat this year, with 57 deals struck in the first five months of 2025, compared with 56 a year earlier, and was concentrated mostly among smaller lenders, according to data from S&P Global Market Intelligence. Major banks seeking selective, or bolt-on acquisitions that add operations such as wealth management, fintech or crypto, will find it easier to get approval from regulators, one of the executives said. But larger mergers involving entire banks that serve similar geographies are more likely to face government scrutiny, including from antitrust authorities, the executive said. Regional lenders are more likely to get the green light for transactions, said Tom Michaud, CEO of investment bank Keefe, Bruyette & Woods. "There is a clear case for gaining scale, and people are realizing this administration gives them the best chance of getting a large deal approved," Michaud said. "So it's better to do it sooner than later," he said, expecting regional lenders to strike deals more quickly than banking giants. The other three industry executives concurred that deals by so-called super regional banks were most feasible. They cited PNC Financial Services (PNC.N), opens new tab, U.S. Bancorp (USB.N), opens new tab and Truist Financial (TFC.N), opens new tab as potential participants. Truist and U.S. Bancorp declined to comment. PNC CEO Bill Demchak said in June that he expected consolidation in retail banking to boost industry profits. Meanwhile, Gunjan Kedia, who became U.S. Bancorp CEO this year, said in February that it was focused on organic growth and ruled out M&A "for now." For the six biggest U.S. lenders deemed by regulators as global systemically important banks, or GSIBs, there are bigger hurdles. JPMorgan Chase (JPM.N), opens new tab and Bank of America (BAC.N), opens new tab, the first and second-largest lenders in the U.S., each hold more than 10% of the nation's deposits and are capped from buying companies that store them. Still, JPMorgan purchased several fintech firms and BofA bought loan portfolios in recent years. Wells Fargo (WFC.N), opens new tab has only recently got out from under key regulatory punishments, while Citigroup (C.N), opens new tab is still under regulators' orders to fix widespread deficiencies in risk management. That leaves Morgan Stanley (MS.N), opens new tab and Goldman Sachs (GS.N), opens new tab as the largest lenders that could pursue the most traditional M&A deals, the three industry executives said. All the six large banks declined comment. The Federal Reserve's new Vice Chair for Supervision, Michelle Bowman, is expected to facilitate deals because of her support for lighter regulation, the three industry executives said. Regulators are generally going to be open to large institutions expanding, but the approval process will remain extensive, said Katie Cox, a consultant CoxFedLaw who previously served as an M&A expert at the Fed. Participants need to show they meet financial and compliance ratings and hold public consultations, Cox said. The process takes at least a year and could probably be sped up to nine months, she added. Regulators would also weigh how combining banks would affect financial stability, and "that's going to be the problem for the G-SIBs -- if the acquisition of any target is going to exacerbate their current financial stability position in the U.S. markets," she said. "And then there's the competition and antitrust rules." Bankers point to a 2023 example as a cautionary tale of the Biden era's skepticism toward deals. After more than a year of waiting for regulatory approvals, Toronto-Dominion Bank ( opens new tab called off its $13.4 billion takeover of First Horizon (FHN.N), opens new tab, triggering a near 40% fall in the latter bank's shares. Industry executives were still watching BNY, which also has GSIB status, to see whether it will continue to pursue Northern Trust or set its sights elsewhere. The approach is being seen as a test case for the administration's openness to GSIB deals, which could reshape the industry because they involve the biggest and most complex institutions, the three executives said.
Reuters
39 minutes ago
- Reuters
Crypto exchanges rushed to list Trump's coin - leaving many losers and some big winners
NEW YORK, July 14 (Reuters) - Crypto exchange Coinbase assures users on its website that it puts any new digital coin through "rigorous" vetting before allowing it to trade. It's an at-times lengthy process meant to protect customers by examining the people connected to the project and the risk of market manipulation or other scams. With President Donald Trump's crypto token, $TRUMP, Coinbase made up its mind in just one day. The $TRUMP token, which launched three days before his inauguration in January, is a meme coin. Based on cultural fads or celebrities, these coins have no intrinsic value and – past experience has shown – are prone to large price swings that can leave investors with losses. A Reuters analysis of crypto market data and industry announcements found that, compared to other recent large meme coins, the biggest crypto exchanges took Trump's to market with unusual speed, despite stating they vet risky coins thoroughly to protect small investors. Some also approved the listing in spite of the high share of coins concentrated in the hands of Trump and his partners, which would normally represent a red flag because of the risk that dumping of tokens by insiders could collapse the price and hurt other investors, some executives said. After reaching an all-time high of $75.35 on April 19, just two days after its launch, $TRUMP crashed to the $7 range by early April, leaving many holders nursing losses. It was trading around $9.55 Thursday. "When the president of the United States launches a meme coin, I thought I might as well put some money inside," said Carl 'Moon' Runefelt, a Dubai-based crypto investor who runs a bitcoin trading channel on YouTube called the "Moon Show." Runefelt said he bought $300,000 worth of the meme coin in tranches at between $50 and $60: "It's probably one of my worst trades, unfortunately." The Reuters analysis showed that eight of the 10 largest crypto exchanges by market share listed the coin within 48 hours of its release. The ninth, Coinbase, added $TRUMP to its listings roadmap on January 18 – indicating it had decided to accept it - and listed the coin three days later. The tenth, Upbit, listed $TRUMP on February 13. That was much faster than they've done on average with the biggest meme coins. Reuters examined how long it took the same 10 exchanges - Binance, Bitget, MEXC, OKX, Coinbase, Bybit, Upbit, and HTX - to list the four other largest meme coins launched since 2022. These, measured by market cap on May 29, are Pepe, Bonk, Fartcoin and dogwifhat. All 10 exchanges listed Pepe and Bonk. Nine listed dogwifhat, and seven listed Fartcoin. On average, the 10 exchanges took 129 days to list those coins. For $TRUMP, they took an average of four. Asked for comment about why they listed $TRUMP so quickly, Bitget, MEXC, OKX, Coinbase and Upbit all said they had not cut any corners with their vetting process. The other five exchanges did not respond to Reuters' questions. Three – Bitget, Coinbase, MEXC – said they moved fast to respond to overwhelming demand for the $TRUMP coin. "The crypto space was buzzing with the hype and, as any other token with a growing craze, it was imperative to add TRUMP," Gracy Chen, Bitget's CEO, said in a statement. Chen said the fact that Trump himself announced the coin on his social media accounts "should kind of solve the compliance issue," citing the fact that "he's the president of the United States." Reuters found no suggestion that Trump or anyone related to his businesses exerted pressure on the exchanges. In response to a request for comment, a White House press official told Reuters the president's assets had been placed in a family trust: "There are no conflicts of interest because the president isn't managing the assets. Any insinuation that there is a conflict of interest is irresponsible." The official referred specific questions about the meme coin to the Trump Organization, which did not respond to Reuters. Coinbase said the $TRUMP token got no special exceptions and the exchange followed its normal process when listing the coin. Paul Grewal, Coinbase's chief legal officer, said many people had to work over the weekend to get the listing done quickly, but no steps were skipped. "Given the information that was shared publicly, we were confident that users could engage with the token positively and safely," Grewal told Reuters. Coinbase listed $TRUMP as an "experimental" token to indicate it comes with "certain risks, including price swings," according to the company's website. The vetting of coins often focuses on how well-known the issuer is, how likely they are to remain in the public eye and how much they engage with the online community to sustain interest in the coin, metrics that $TRUMP would score highly on, according to Santa Clara University finance professor Seoyoung Kim, who specializes in crypto analytics. She cautioned that focusing on vetting speed alone could provide an incomplete picture of investor protection. A more holistic analysis, Kim said, would also involve factors such as the average market cap at which a coin is listed, for how long it has sustained that level before its listing, and its daily trading volumes. With $TRUMP listed so soon after launch, there was little such data for exchanges to parse. $TRUMP's market cap has since fallen to around $1.9 billion, down sharply from its peak above $15 billion on January 19. But that still ranks it amongst the largest meme coins launched since 2022. Reuters ran its listing-speed analysis past five academics with crypto expertise, including Kim, who all said its methodology was sound. David Krause, Emeritus Professor of Finance at Marquette University, who has studied Trump's crypto ventures, said the quickness of the $TRUMP listing "suggests either a dramatic acceleration of due diligence or corners being cut." "Either scenario has significant implications for investor protection and market integrity," he said. The president's rush of business ventures in a lightly-regulated sector that his government is responsible for overseeing has drawn criticism from Democrats, consumer advocacy groups and former financial enforcement officials. "You don't say no to hosting the president's new meme coin," said Corey Frayer, a former senior crypto advisor at the U.S. Securities and Exchange Commission. Frayer is now director of a non-profit advocacy group, the Consumer Federation of America. "The president controls who oversees your business and how they enforce the law." Under former President Joe Biden, the SEC maintained that most crypto tokens, including meme coins, should be regulated as securities, making exchanges cautious about listing them. That began to change, quickly, after Trump was elected last November. The Republican has styled himself as the "crypto president," pledging to overhaul regulation of the sector. Following Trump's election, Coinbase – the largest publicly traded crypto exchange in the United States – and several of its rivals began listing more meme coins. In Trump's second term, the SEC has paused or withdrawn high-profile enforcement actions against crypto operators, including a major investor in a Trump family crypto project, and issued a staff statement concluding that meme coins do not constitute securities. An SEC spokesperson declined to comment on the agency's crypto policy and Trump's coin. Trump's family has launched multiple crypto ventures, raking in hundreds of millions of dollars. The $TRUMP token quickly earned an estimated $320 million in fees, though it's not publicly known how that amount has been divided between a Trump-controlled entity and its partners. Exchanges have been major beneficiaries of Trump's embrace of the industry. $TRUMP has generated significant revenue for the 10 exchanges in Reuters' review: more than $172 million in trading fees, according to estimates based on standard fees compiled for the news agency by CoinDesk Data, a crypto industry data provider. Trade in the coin, meanwhile, has favored a small group of investors. At the top, 45 crypto wallets cleared about $1.2 billion in profits overall, while another 712,777 wallets have collectively lost $4.3 billion, according to trading data analyzed by crypto analysis firm Bubblemaps as of June 18. In the middle, more than half a million wallets made an average of $5,656 profit each. In listing $TRUMP, some exchanges proceeded despite a factor they'd previously labelled as a red flag: 80% of the coin's supply was held by the Trump family and its partners. Such a high concentration of ownership can allow the team behind a coin to sell large amounts of it at once, collapsing the price for retail investors. The terms of the $TRUMP coin specified that its total supply would be gradually unlocked over three years after initial release. On January 16, the day before $TRUMP was released, the New York State Department of Financial Services issued an alert to consumers about the risks of meme coins. Such coins, the notice said, are carried by platforms not licensed by the state and the supply of the digital tokens is often controlled by a small number of people. That opens the door to "pump-and-dump schemes," the regulator noted, in which public hype by their issuers leads to a jump in price – with big, early investors exiting and smaller retail buyers left holding the losses that follow. The NYDFS declined to comment beyond the guidance. Coinbase, which is subject to New York regulations, blocked state residents from accessing the token, but allowed U.S. customers elsewhere to trade. To list $TRUMP in New York, the exchange would have faced a long list of risk assessment and governance requirements. Some other exchanges acknowledged they looked past concerns about the concentration in a bid to serve customer demand. MEXC's chief operating officer, Tracy Jin, told Reuters that, because of the concentration of tokens, $TRUMP did not meet its usual standards for a full listing on its main board, but the exchange pushed ahead anyway due to strong demand. In a follow-up written statement, an MEXC spokesperson said that a "faster-than-usual" listing was possible because the coin had clear market momentum and it met "our listing standards early." Commenting on the Reuters listing-speed analysis, the spokesperson said market conditions and demand for political meme tokens had changed since 2022, "making direct comparisons less relevant." Bitget also had concerns about the 80% figure, CEO Chen told Reuters. "Eighty percent held by the team, even though there's a little bit of a lock-up period, is in my opinion very risky," said Chen. "Ultimately, user trading volume, demand … overrode the so-called risky factor here." Like some exchanges, Bitget, based in the Seychelles, does not have a business presence in the U.S. or serve clients who reside there, Chen said. "Globally," she added, "people are generally aware of the risks associated with trading meme coins." Upbit, which operates in South Korea, said it does not comment on specific coin listings but that it has "a rigorous and comprehensive evaluation process." Erald Ghoos, CEO for Europe of OKX, said the exchange's legal and compliance teams stayed up all night over different time zones to work on the listing. Seychelles-registered OKX says its diligence process requires "meticulous preparation." It decided to list $TRUMP within 26 hours.
The Independent
43 minutes ago
- The Independent
US manufacturers are stuck in a rut despite subsidies from Biden and protection from Trump
Democrats and Republicans don't agree on much, but they share a conviction that the government should help American manufacturers, one way or another. Democratic President Joe Biden handed out subsidies to chipmakers and electric vehicle manufacturers. Republican President Donald Trump is building a wall of import taxes — tariffs — around the U.S. economy to protect domestic industry from foreign competition. Yet American manufacturing has been stuck in a rut for nearly three years. And it remains to be seen whether the trend will reverse itself. The U.S. Labor Department reports that American factories shed 7,000 jobs in June for the second month in a row. Manufacturing employment is on track to drop for the third straight year. The Institute for Supply Management, an association of purchasing managers, reported that manufacturing activity in the United States shrank in June for the fourth straight month. In fact, U.S. factories have been in decline for 30 of the 32 months since October 2022, according to ISM. 'The past three years have been a real slog for manufacturing,'' said Eric Hagopian, CEO of Pilot Precision Products, a maker of industrial cutting tools in South Deerfield, Massachusetts. 'We didn't get destroyed like we did in the recession of 2008. But we've been in this stagnant, sort of stationary environment.'' Big economic factors contributed to the slowdown: A surge in inflation, arising from the unexpectedly strong economic recovery from COVID-19, raised factory expenses and prompted the Federal Reserve to raise interest rates 11 times in 2022 and 2023. The higher borrowing costs added to the strain. Government policy was meant to help. Biden's tax incentives for semiconductor and clean energy production triggered a factory-building boom – investment in manufacturing facilities more than tripled from April 2021 through October 2024 – that seemed to herald a coming surge in factory production and hiring. Eventually anyway. But the factory investment spree has faded as the incoming Trump administration launched trade wars and, working with Congress, ended Biden's subsidies for green energy. Now, predicts Mark Zandi, chief economist at Moody's Analytics, 'manufacturing production will continue to flatline.' 'If production is flat, that suggests manufacturing employment will continue to slide,' Zandi said. 'Manufacturing is likely to suffer a recession in the coming year.'' Meanwhile, Trump is attempting to protect U.S. manufacturers — and to coax factories to relocate and produce in America — by imposing tariffs on goods made overseas. He slapped 50% taxes on steel and aluminum, 25% on autos and auto parts, 10% on many other imports. In some ways, Trump's tariffs can give U.S. factories an edge. Chris Zuzick, vice president at Waukesha Metal Products, said the Sussex, Wisconsin-based manufacturer is facing stiff competition for a big contract in Texas. A foreign company offers much lower prices. But 'when you throw the tariff on, it gets us closer,'' Zuzick said. 'So that's definitely a situation where it's beneficial.'' But American factories import and use foreign products, too – machinery, chemicals, raw materials like steel and aluminum. Taxing those inputs can drive up costs and make U.S producers less competitive in world markets. Consider steel. Trump's tariffs don't just make imported steel more expensive. By putting the foreign competition at a disadvantage, the tariffs allow U.S. steelmakers to raise prices – and they have. U.S.-made steel was priced at $960 per metric ton as of June 23, more than double the world export price of $440 per ton, according to industry monitor SteelBenchmarker. In fact, U.S. steel prices are so high that Pilot Precision Products has continued to buy the steel it needs from suppliers in Austria and France — and pay Trump's tariff. Trump has also created considerable uncertainty by repeatedly tweaking and rescheduling his tariffs. Just before new import taxes were set to take effect on dozens of countries on July 9, for example, the president pushed the deadline back to Aug. 1 to allow more time for negotiation with U.S. trading partners. The flipflops have left factories, suppliers and customers bewildered about where things stand. Manufacturers voiced their complaints in the ISM survey: 'Customers do not want to make commitments in the wake of massive tariff uncertainty,'' a fabricated metal products company said. 'Tariffs continue to cause confusion and uncertainty for long-term procurement decisions,'' added a computer and electronics firm. 'The situation remains too volatile to firmly put such plans into place.'' Some may argue that things aren't necessarily bad for U.S. manufacturing; they've just returned to normal after a pandemic-related bust and boom. Factories slashed nearly 1.4 million jobs in March and April 2020 when COVID-19 forced many businesses to shut down and Americans to stay home. Then a funny thing happened: American consumers, cooped up and flush with COVID relief checks from the government, went on a spending spree, snapping up manufactured goods like air fryers, patio furniture and exercise machines. Suddenly, factories were scrambling to keep up. They brought back the workers they laid off – and then some. Factories added 379,000 jobs in 2021 — the most since 1994 — and then tacked on another 357,000 in 2022. But in 2023, factory hiring stopped growing and began backtracking as the economy returned to something closer to the pre-pandemic normal. In the end, it was a wash. Factory payrolls last month came to 12.75 million, almost exactly where they stood in February 2020 (12.74 million) just before COVID slammed the economy. 'It's a long, strange trip to get back to where we started,'' said Jared Bernstein, chair of Biden's White House Council of Economic Advisers. Zuzick at Waukesha Metal Products said that it will take time to see if Trump's tariffs succeed in bringing factories back to America. 'The fact is that manufacturing doesn't turn on a dime,'' he said. 'It takes time to switch gears.'' Hagopian at Pilot Precision is hopeful that tax breaks in Trump's One Big Beautiful Bill will help American manufacturing regain momentum. 'There may be light at the end of the tunnel that may not be a locomotive bearing down,'' he said. For now, manufacturers are likely to delay big decisions on investing or bringing on new workers until they see where Trump's tariffs settle and what impact they have on the economy, said Ned Hill, professor emeritus in economic development at Ohio State University. 'With all this uncertainty about what the rest of the year is going to look like,'' he said, 'there's a hesitancy to hire people just to lay them off in the near future.'' 'Everyone,'' said Zuzick at Waukesha Metal Products, 'is kind of just waiting for the new normal.''
