AI companies and their customers are paying increasing attention to the cost of running large-scale deployments. While open-source models can reduce the price per token, businesses still face the challenge of choosing the right model and making it efficient for different workloads.

On Thursday, Writer introduced its new flagship AI model, Palmyra X6, alongside upgrades to its agentic harness, the infrastructure that helps AI agents complete tasks. Writer says the combination could reduce customer costs by as much as 50% for basic workloads.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

A Focus On Lower-Cost AI

Palmyra X6 is built as a post-training version of Z.ai’s open-source GLM-5.2 model. Writer says the system is designed to deliver deployment-ready capabilities while using fewer tokens and completing complex, multi-step tasks more efficiently.

Writer CEO May Habib told TechCrunch that enterprise customers are becoming less interested in chasing benchmark records and more focused on controlling the cost of AI deployments.

The company’s new model and harness upgrades are available to Writer customers starting Thursday.

Why The AI Harness Matters

Writer’s approach goes beyond the model itself. The company has also significantly upgraded its standard agentic harness, arguing that improvements to the way AI agents operate can have a major impact on overall costs.

A recent research paper from Writer researchers supports that argument. After testing harness efficiency across multiple models, the researchers found that optimizing the harness was often a more reliable way to reduce costs than simply changing models. Their tests showed an average cost reduction of about 40%.

The researchers described the harness as a component whose efficiency can multiply across every model an organization uses, making optimization potentially valuable even as companies switch between models.

Keeping The Model Choice Flexible

Palmyra X6 will not replace other models available through Writer. Customers can continue using Writer’s models alongside third-party systems imported through Microsoft Azure or Amazon Bedrock.

Habib also sees the growing focus on cost efficiency as a sign that enterprises are becoming more skeptical of major AI labs. She argued that companies are increasingly concerned about the rising cost of AI deployments and whether model providers are sufficiently focused on helping businesses generate practical value.

For Writer, the strategy is therefore not simply about launching another AI model. By combining a lower-cost system with a more efficient agentic infrastructure, the company is positioning cost control as a central part of enterprise AI adoption.

Surveillance technology company Flock Safety has introduced new policies and tools designed to reduce misuse of its automated license plate reader systems and increase accountability among law enforcement customers.

The changes include shorter data retention periods, stricter rules for sharing information and a monitoring tool called Audit Assistance. However, Flock has provided limited information about how the system actually identifies suspicious activity.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

New Rules For Data Retention

Flock is now recommending that police departments retain license plate data for seven days, down from 30 days. For exceptional cases requiring longer retention, customers can use “Evidence Mode,” which requires a specific case number.

Departments will also be able to restrict data sharing with other Flock customers based on the type of offense being investigated. The company announced these measures alongside its new privacy and transparency policies.

How Audit Assistance Works

Flock introduced Audit Assistance in April and said more than one-third of its customers have activated it. The company now plans to require all customers to enable the tool by the end of the year. According to Flock, Audit Assistance can detect unusual search patterns and flag them for administrator review. Users may also be automatically locked out until an administrator investigates the activity.

Company executives have offered some examples of what the system can identify. Flock’s head of trust and compliance Ashley Haber said it can flag an unusual search history, while co-founder Paige Todd pointed to cases where a user repeatedly searches for the same license plate over an extended period.

Flock describes the feature as a tool that can “surface atypical activity early” and provide a documented process for reviewing it. The company says it is not based on machine learning or AI, but instead analyzes search patterns. One example that could trigger a flag is searching for the same license plate under multiple case codes, which could indicate potential misuse.

However, a flag does not necessarily mean abuse occurred. It simply indicates that further investigation may be warranted.

Questions About Effectiveness Remain

Flock has not disclosed detailed information about the system’s effectiveness, including its false-positive and false-negative rates or whether an independent organization has audited it.

That lack of transparency has drawn criticism from privacy advocates. Chad Marlow, senior policy counsel at the American Civil Liberties Union, argued that there is not enough evidence to determine how consistently the tool detects abuse. He said independent testing would be necessary to establish whether it is a meaningful security measure or merely “window dressing.”

Cooper Quintin, a security researcher at the Electronic Frontier Foundation, also questioned whether the tool alone can prevent misuse. In his view, accountability ultimately depends on whether officers who abuse the technology face meaningful consequences.

Rachel Levinson-Waldman of the Brennan Center for Justice described mandatory use of Audit Assistance as a positive step, but said the system itself should be independently audited because too little is known about how it performs in practice.

Flock Faces Continued Scrutiny

The announcement comes after years of reports involving law enforcement officers allegedly misusing Flock’s technology. Earlier this week, three former Georgia sheriff’s deputies were arrested over allegations that they used Flock cameras to monitor people with whom they had personal relationships.

The Bibb County Sheriff’s Office said Flock’s new Audit Assistance tool helped uncover the alleged misuse, giving the company an example of the technology working as intended.

Still, questions remain over how reliably the system can detect abuse and whether law enforcement agencies will be able to hold users accountable when violations occur.

What began as a major success for India’s startup ecosystem has turned into a complex legal dispute less than a year after VideoVerse was acquired for $250 million.

The deal was announced in September 2025 by VideoVerse and international sports publisher Minute Media. VideoVerse had developed AI-powered software for turning sports broadcasts into short clips, with plans to expand the technology internationally.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

The deal has since unravelled. Investors are still waiting for proceeds, while founder Vinayak Shrivastav faces multiple legal claims. In May, Minute Media terminated its agreement with VideoVerse, citing “significant discrepancies” in the company’s representations.

Investors Seek Millions

Bluestone Capital, which backed VideoVerse in 2023, is suing the company for fraud and alleges that it failed to distribute acquisition proceeds as required.

Another creditor is seeking $64 million from a loan Shrivastav took out shortly after the acquisition. The complaint alleges that fraudulent merger documents were used to secure shareholder approval.

Former COO Sabya Das has separately accused Shrivastav of forging his signature on loan and share-repurchase agreements that allegedly resulted in tens of millions of dollars being extracted from the company.

The allegations have not been proven in court, and Shrivastav did not respond to requests for comment.

Loan Raises Further Questions

In October 2025, Shrivastav arranged a $55 million structured loan from investment firm Lingotto. According to court filings, $53 million was transferred to an account controlled by VideoVerse.

Lingotto now alleges that documents supporting the loan were forged, including papers supposedly signed by Minute Media’s CEO, while screenshots showing company bank balances were also allegedly fabricated.

After a $4 million payment due in March was missed, Lingotto demanded repayment and discovered other creditors were also awaiting payments. Shrivastav was removed as CEO by the end of April.

From AI Startup To Legal Dispute

VideoVerse had built a strong position in automated sports content through its Magnifi platform, which uses AI to identify key moments and players and create short-form clips. Its customers included the Indian Premier League, FIFA+ and Nippon TV.

Minute Media had hoped to use the technology to expand internationally. Instead, the acquisition has triggered multiple legal battles over missing funds, disputed agreements and the conduct of the company’s leadership.

Cases involving Minute Media, Lingotto, Bluestone Capital and former executives are now being heard in Delaware Chancery Court, leaving investors and creditors seeking answers about what happened to the money and whether the $250 million deal received adequate due diligence.

Cyprus is looking to make artificial intelligence a key part of its tourism strategy as the government explores ways to improve services, enhance the visitor experience and strengthen the island’s competitiveness.

Tourism Deputy Minister Kostas Koumis and Chief Scientist for Research, Innovation and Technology Demetris Skourides discussed the plans on Thursday, with the meeting focusing on how AI can support the digital transformation of Cyprus as a tourism destination.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

Smart Tourism Strategy Takes Shape

Skourides’ team presented “Smart Tourism 2032: the Cyprus Artificial Intelligence Strategy” as part of the ongoing public consultation on the country’s national AI strategy.

The consultation is due to close on August 31, 2026. The tourism initiative forms part of the wider effort to determine how AI should be developed and applied across the economy.

Under the proposed strategy, Cyprus would become a “Living Lab” combining advanced technology with its traditional hospitality. The approach is built around three areas: smart tourism infrastructure, smart destination planning and management, and an empowered digital tourist.

Small and medium-sized tourism businesses are also expected to play an important role in adopting new technologies across the sector.

AI To Reshape The Visitor Experience

According to Koumis, AI is already changing tourism, from how visitors research destinations and plan trips to how destinations promote themselves.

“AI has entered the tourism sector dynamically and has already brought visible changes to a series of tourism-related functions, such as information gathering, trip planning and destination promotion,” he said.

The government wants to work with Skourides’ team to use the technology both to improve Cyprus’ competitiveness and enhance the experience offered to visitors.

Koumis said the development of tourism services was increasingly connected to the new capabilities created by AI, adding that Cyprus should make the most of these opportunities.

Tourism Businesses Included In AI Plans

Beyond improving individual services, the strategy aims to give tourism businesses a role in shaping the country’s broader AI framework.

The Deputy Ministry of Tourism wants the sector to participate in the consultation and take advantage of support created through technological development and the expansion of AI.

Ultimately, “Smart Tourism 2032” seeks to combine digital innovation with Cyprus’ established tourism strengths, particularly its reputation for hospitality, while preparing the sector for a more technology-driven visitor journey.

Cyprus banks continued to report relatively strong asset quality in May, with the non-performing loan (NPL) ratio holding steady at 1.6%, according to the Central Bank of Cyprus (CBC).

The figure remained unchanged from April and was below the EU-wide NPL ratio of 1.98% recorded in March, based on consolidated banking data from the European Central Bank (ECB). While the reporting dates differ, Cyprus’ ratio was 0.38 percentage points lower than the EU average.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

NPL Coverage Edges Higher

Banks also slightly strengthened their protection against potential losses. The NPL coverage ratio increased to 63.0% at the end of May, from 62.9% a month earlier, meaning that provisions covered nearly two-thirds of non-performing loans.

Meanwhile, the stock of restructured loans remained relatively contained. Total restructured loans stood at €800 million, of which €300 million were still classified as non-performing.

Cyprus Banks Show Strong Profitability

Cyprus also compared favourably with the wider EU banking sector on profitability during the first quarter of 2026.

Domestic banks recorded a return on equity of 3.6973%, compared with 2.44% across EU credit institutions in March. The measure indicates how effectively banks generate profits from shareholders’ capital.

Across the EU, banks continued to maintain substantial capital buffers. The Common Equity Tier 1 ratio stood at 16.27% in March, providing an important cushion against potential losses.

EU Banking Sector Remains Resilient

The ECB’s March data covered 335 banking groups and 2,284 stand-alone credit institutions, alongside non-EU-controlled subsidiaries and branches operating within the bloc. Together, these institutions represented almost the entire EU banking sector by balance sheet.

Aggregate assets of EU-headquartered credit institutions rose 3.63% year on year, reaching €34.33 trillion in March 2026, up from €33.13 trillion a year earlier.

Overall, Cyprus’ latest figures point to a banking sector with relatively contained asset-quality pressures, supported by a stable 1.6% NPL ratio, 63% coverage and profitability above the EU average. The comparison should be viewed with some caution, however, as the CBC and ECB figures cover different reporting periods and datasets.

SK Hynix is investing $720 billion in what it says will become the world’s largest network of memory factories, betting that demand for AI chips will remain strong for years to come.

The South Korean memory giant, whose market value has climbed more than fivefold over the past year to above $1 trillion, is expanding production as AI companies compete for limited supplies of high-bandwidth memory (HBM).

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

AI Drives A Memory Race

HBM is essential for AI processors because it enables rapid data access. SK Hynix held 58% of the global HBM market in the first quarter, ahead of Samsung and Micron, which each had 21%, according to Counterpoint Research.

Demand has pushed memory prices higher and encouraged major technology companies to secure supply through long-term agreements. SK Hynix signed 10 such deals in July, while Nvidia agreed to secure HBM supply and co-develop next-generation memory as part of a broader $500 billion deal with SK Group.

“It’s like a war,” said Chey Tae-won, chairman of SK Group, which controls SK Hynix. “Everybody wants to buy the memory chips.”

Nvidia CEO Jensen Huang has even sent SK Hynix a message on a wafer: “Please make more.”

Building A New Memory Hub

At the centre of SK Hynix’s expansion is the Yongin Cluster, where the company is building four fabs. The first will rise to roughly the height of a 50-story apartment building and feature six cleanrooms across multiple floors.

The company is also expanding its facilities in Cheongju, while South Korea is pursuing a broader plan to double national memory production over the next five years.

SK Hynix is not alone in the race. Micron is investing $50 billion in two fabs in Idaho and plans a potential $100 billion campus in New York. The Korean company is also building a $4 billion packaging facility in Indiana, scheduled for completion in 2028.

China Adds Pressure

Alongside the global race for capacity, SK Hynix faces growing competition from China. The company operates three fabs there but cannot sell its most advanced HBM products in the country because of U.S. export controls.

Chinese memory maker CXMT is expanding rapidly and recently made a high-profile debut on the Shanghai stock market. “It’s a race, and now the counterparty of the race is China,” Counterpoint Research director MS Hwang said.

For SK Hynix, the next stage of growth will increasingly depend on custom HBM designed specifically for AI processors. The company believes this shift could make memory less of a commodity and help protect its massive investment.

“Nvidia wants their own custom chips and Google wants their own customized HBM, so it’s not just a commodity,” Tae-won said. “It actually changes the memory chip’s status.”

Databricks has closed a $5 billion funding round at a $190 billion valuation, marking a significant increase from the $134 billion valuation it reached just six months ago. The company said Thursday that its revenue run rate surpassed $7 billion in the second quarter, with revenue growing more than 80% year over year.

Funding To Accelerate Enterprise AI

Databricks plans to use the new capital to expand its enterprise AI capabilities, including its Unity AI Gateway governance platform and Genie agentic tools.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

Founded in 2013, Databricks helps businesses build AI applications and agents using their proprietary data. The latest round comes after the company raised $5 billion and secured $2 billion in additional debt capacity earlier this year.

Expanding Beyond Data Analytics

The company has been moving beyond its core data platform into several new areas. Its recently launched Lakebase database, which competes with companies such as Oracle and SAP, has already surpassed a $100 million revenue run rate, according to Databricks.

Meanwhile, its Lakehouse data warehousing business has exceeded a $1.5 billion run rate, while Lakewatch marked the company’s entry into cybersecurity earlier this year.

Databricks ranked No. 3 on CNBC’s 2026 Disruptor 50 list and has grown into a major private-market competitor to Snowflake.

Private Markets Keep IPO Pressure Low

Large private funding rounds are allowing companies such as Databricks to delay going public. Meanwhile, Anthropic and OpenAI are preparing for potential IPOs, highlighting the growing competition for investor capital across private and public AI companies.

The latest Databricks round was led by Coatue, Blackstone, MGX, T. Rowe Price and Sixth Street Growth.

Classic car auctions during Monterey Car Week could generate as much as $500 million this year, potentially surpassing the previous record as younger collectors drive demand for modern supercars.

Hagerty estimates total sales of $470 million to $500 million, which would exceed the $471 million record set in 2022 and extend the market’s recovery after weaker results in 2023 and 2024. “With strong bidding, this could be the first half-billion-dollar auction week the collector world has ever seen,” said McKeel Hagerty, CEO of Hagerty.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

Younger Collectors Reshape The Market

A generational shift is changing what buyers want. Millennials and Gen Z collectors are increasingly turning to the supercars they grew up admiring rather than the classic models that dominated the market for decades.

Ferrari F40s, F50s and Enzos, along with Bugatti Veyrons, Koenigseggs and Paganis, have recorded sharp price gains, with some models doubling in value over the past two years.

Among Monterey’s biggest lots is a 1996 McLaren F1 GTR, estimated at $35 million by RM Sotheby’s. A 2023 Ferrari Daytona SP3 could also rank among the top 10, with an estimate above $10 million.

Supercars Outpace Traditional Classics

The shift is reflected in Hagerty’s indexes. Its Blue Chip Index, which tracks leading traditional collector cars, fell 2% over the past year, while the Supercar Index climbed 30%.

Rapid appreciation has raised concerns about speculation, with some dealers arguing that prices for modern supercars are increasingly disconnected from traditional measures such as rarity, racing history and long-term collectability.

“There is a huge amount of speculation in that part of the market,” said classic car dealer and adviser Simon Kidston, describing the market as “very frothy.”

Recent sales highlight the trend. A 2003 Ferrari Enzo sold for $17.9 million in January, nearly three times its previous auction record, while another Enzo reached $15.2 million in March. A 2005 Porsche Carrera GT sold for $6.7 million, more than doubling its previous record.

Ferrari Still Leads The Market

Ferrari remains dominant at the top end of the collector market. Nine of the 10 most expensive cars sold at auction so far this year have been Ferraris, according to Hagerty, while five of Monterey’s top lots come from the Italian marque.

“All roads lead to Maranello,” Hagerty said.

While Ferraris from the 1980s, 1990s and early 2000s are gaining value, many celebrated models from the 1950s and 1960s have largely stalled.

Younger buyers are not exclusively chasing modern cars. Kidston recently sold a 1967 Ferrari 275 GTB/4 to a 35-year-old technology founder who called it his dream car, showing that a new generation is also developing an interest in classic models.

China’s Yangtze Memory Technologies (YMTC) is gaining ground in the global NAND memory market, reaching third place by shipments in the second quarter, according to Counterpoint Research.

YMTC captured 14% of global NAND shipments, putting it behind South Korea’s Samsung and SK hynix but ahead of U.S. chipmaker Micron and Japan’s Kioxia.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

YMTC Strengthens Its Position

NAND memory retains data when devices are powered off and is widely used in smartphones, computers and storage products. Demand is rising as the need for data storage grows, although NAND is generally slower and less expensive than DRAM.

Counterpoint Research Director MS Hwang expects YMTC to extend its lead over Kioxia in 2027 and 2028. He also said manufacturers need roughly 15% market share to generate enough cash to fund future capital investment.

YMTC had briefly overtaken Kioxia a year ago before falling behind again. Its return to third place therefore marks an important shift in the competitive landscape.

China’s Memory Sector Expands

YMTC is preparing for a potential listing in mainland China, following the successful debut of Chinese DRAM maker CXMT in July.

CXMT held 7% of the global DRAM market in the second quarter, placing fourth behind Samsung, SK hynix and Micron, according to Counterpoint’s DRAM report.

Despite its shipment gains, YMTC still trails Micron and Kioxia in NAND revenue because its business remains more focused on consumer products than data centers. Counterpoint expects data centers to account for around half of NAND demand by the end of 2026.

Meanwhile, SK hynix is reportedly preparing to resume investment at its Dalian facility in China after a four-year pause, potentially boosting production capacity.

Oil prices fell on Thursday as investors balanced weaker global demand expectations against ongoing disruptions to energy supplies in the Middle East.

U.S. West Texas Intermediate futures declined about 2% to $81.61 a barrel, while Brent crude, the international benchmark, was down 1.8% at $87.40.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

Demand Outlook Weakens

The International Energy Agency said global oil demand is now expected to decline more sharply than previously forecast, with the prolonged closure of the Strait of Hormuz weighing increasingly on the market.

At the same time, efforts to restore global supply remain constrained by renewed hostilities and disruptions to maritime transport. Global oil supply was still 6.3 million barrels per day below year-earlier levels in July, according to the IEA.

Middle East Risks Remain

Security concerns continue to affect shipping routes across the region, with attacks reported on vessels in the Gulf of Oman and the Red Sea this week.

Oman is also dealing with an oil spill after a tanker carrying an estimated 800,000 barrels of Russian crude ran aground on June 30. The vessel has reportedly begun leaking oil along the country’s coastline, including near a nature reserve home to Arabian Sea humpback whales and Socotra cormorants. Reuters report on the Oman oil spill.

Diplomatic efforts to reopen the Strait of Hormuz appear to be continuing, but the conflict, now more than five months old, continues to disrupt regional energy flows.

“The lack of clarity over the possibility of a full reopening of the waterway could leave oil prices exposed to the upside at a time when the market remains tight,” said Christopher Tahir, senior market strategist at Exness. He added that further setbacks could push prices higher, leaving the oil market caught between deteriorating demand prospects and persistent supply risks.