Alibaba Bets on 10 Trillion-Parameter AI Model and Homegrown Chips in China’s Race to Build an AI Stack

By

-

September 22, 2026

Alibaba is escalating its push to compete in artificial intelligence, announcing plans for an AI model potentially four times larger than its current flagship while unveiling a new processor it described as China’s most powerful AI chip.

The announcements, made Tuesday at Alibaba Cloud’s annual Apsara conference in Hangzhou, sent Alibaba’s Hong Kong-listed shares up 5.1% to their highest level in a month.

The moves highlight how Alibaba is attempting to build a vertically integrated AI business spanning models, semiconductors, cloud computing and the data-center infrastructure needed to train and operate powerful AI systems.

The strategy is also taking shape as Chinese technology companies accelerate efforts to develop domestic alternatives to Nvidia’s AI processors. U.S. export restrictions have limited Chinese companies’ access to some of the world’s most advanced AI chips, increasing the commercial and strategic importance of domestic semiconductor development.

Alibaba CEO Eddie Wu said the most significant products of what he called the “Machine Intelligence era” have yet to arrive, comparing the potential transformation with the Industrial Revolution.

Wu predicted that machines could eventually generate more than 1,000 times the amount of “thinking” produced by humanity, compared with less than 3% today.

Alibaba Targets 10 Trillion Parameters

The company’s most ambitious announcement concerned the next generation of its Qwen AI models.

Alibaba said its Qwen team plans to develop models with between 5 trillion and 10 trillion parameters, aimed at handling more complex and longer-horizon tasks as the company pursues what Wu described as artificial superintelligence, or systems that surpass human capabilities.

Alibaba’s current flagship, Qwen 3.8 Max, has 2.4 trillion parameters. Parameters are a rough measure of the scale of an AI model, although a larger parameter count does not automatically translate into better performance.

The company said it is currently training Qwen 4, with future Qwen 4.5 and Qwen 5 models expected to scale toward the 5 trillion-to-10 trillion range.

The more significant development may be the degree to which Alibaba says its models are beginning to contribute to their own improvement.

Wu said the Qwen team has made “meaningful” progress in enabling models to identify weaknesses, conduct experiments and generate training data with limited human involvement.

That points toward a more automated model-development cycle in which AI systems increasingly assist with the work required to build the next generation of AI systems. If that process becomes reliable at scale, the constraint on AI development could shift from human engineering capacity toward computing power, energy and access to advanced semiconductor technology.

Analysts believe it is what makes Alibaba’s parallel investment in chips and data centers particularly important.

Building A Domestic Nvidia Alternative

Alibaba also unveiled the Zhenwu V900, a new AI processor developed by its T-Head semiconductor division.

Wu said the chip delivers three times the performance of its predecessor, the M890, and can be connected in clusters of up to 500,000 processors to train and run large AI models.

The chip is scheduled to enter mass production and commercial release in the first quarter of 2027. Alibaba expects AI chip shipments to grow significantly each year.

The development is considered a very big boost because China’s AI ambitions increasingly depend on whether domestic companies can secure enough computing capacity without relying on Nvidia.

China has a number of domestic AI chip developers, but matching the performance, software ecosystem and availability of Nvidia’s products remains a significant challenge. Alibaba’s decision to develop its own processors gives the company another way to control a critical component of its AI infrastructure. It also creates a potential feedback loop across Alibaba’s businesses. The company can develop AI models through Qwen, manufacture or source processors through its semiconductor operations, and deploy those systems through Alibaba Cloud.

That vertical integration could become more valuable if access to foreign AI processors remains restricted. But producing a powerful processor is only part of the challenge. AI chips need software ecosystems, networking technology, data-center infrastructure and large-scale customers to become commercially important.

Alibaba is therefore expanding aggressively in the infrastructure layer as well.

Data Centers Become The Next Bottleneck

Wu said Alibaba Cloud plans to increase its global data-center capacity to more than 20 gigawatts by 2032. That target underscores the scale of the computing infrastructure required to support the company’s AI ambitions.

Alibaba said customer demand for AI remains “exceptionally robust” and is accelerating revenue growth at Alibaba Cloud. But the company is also encountering supply-chain constraints that are limiting how quickly it can expand capacity.

“The industry’s mid-to-long-term demand far outpaces our supply capabilities,” Wu said.

Alibaba Cloud plans to begin bringing its AI supernodes online at commercial scale this quarter.

The imbalance between demand and available computing capacity is becoming a central feature of the AI infrastructure market. For companies such as Alibaba, the challenge is no longer simply developing a competitive model. They also need enough chips, electricity, data centers, and networking equipment to make those models commercially available at scale.

That helps explain why Alibaba is simultaneously investing across the AI stack.

The company’s announcements also provide a broader indication of where China’s AI industry may be heading. Rather than competing only at the application or model layer, Chinese technology companies are increasingly trying to build domestic alternatives across the entire computing chain.

For Alibaba, the commercial opportunity is potentially substantial if Qwen can become a major enterprise AI platform and Alibaba Cloud can capture the resulting demand for computing.

But the investment requirements will also be considerable.

A 20-gigawatt data-center footprint represents a massive long-term infrastructure commitment, while the development and production of increasingly advanced AI chips requires sustained investment in semiconductor engineering and manufacturing.

Alibaba’s ability to execute across those layers will therefore matter as much as the headline parameter counts. Wu compared AI coding with the light bulb during the electrical age, suggesting that today’s widely used applications may represent only an early stage of a much larger technological transformation.

But the more immediate test for Alibaba, like other companies, is whether its growing AI ecosystem can turn that vision into revenue.

Bitcoin’s $87,000 Rally Meets a New Oil-Market Signal

By

Paul Ugbede Godwin

-

September 22, 2026

0

Bitcoin’s move toward $87,000 is arriving at an unusual moment for global markets: geopolitical tensions that recently threatened to push energy prices higher are suddenly showing signs of easing.

On September 21, U.S. spot Bitcoin ETFs attracted roughly $999 million in net inflows, their strongest daily haul of 2026, while Bitcoin briefly climbed above $87,000.

At the same time, Brent crude fell below $100 as investors reacted to developments that could restore oil supply through the Middle East.

The Bitcoin rally is particularly significant because it combines institutional buying with forced positioning in the derivatives market. The near-$1 billion ETF inflow followed a period in which Bitcoin had struggled around lower levels, suggesting that institutional demand was returning through regulated investment vehicles.

BlackRock’s IBIT accounted for about $381 million of the inflow, while ARKB and Fidelity’s FBTC also recorded substantial purchases. At the same time, more than $1 billion in crypto positions were reportedly liquidated during the market’s sharp move.

A large portion of the liquidations involved traders positioned against the rally, creating a classic short squeeze in which rising prices force bearish traders to close positions, adding further buying pressure. Bitcoin briefly reached around $87,300 before retreating toward the mid-$80,000s.

That distinction matters. ETF inflows represent new capital entering Bitcoin investment products, while liquidations are largely a consequence of leverage being unwound. Together, however, they can create an unusually powerful market structure: institutional demand provides the foundation while leveraged positioning accelerates the move.

The oil market is sending a different but connected signal. Iran has indicated that it is prepared to reopen the strategically important Strait of Hormuz within seven days if the United States eases military pressure and lifts its blockade on Iranian ports.

The proposal remains conditional, meaning the reopening is not guaranteed, but the possibility alone has changed expectations around the immediate supply outlook.

Saudi Arabia has also restarted its East-West oil pipeline after a shutdown caused by a drone attack. The pipeline provides an alternative route to the Red Sea port of Yanbu, allowing Saudi crude to bypass the Strait of Hormuz.

Reuters reported that pumping had resumed at a reduced rate, although restoring full capacity could take six to eight weeks because several pumping stations were damaged. The result was immediate in oil markets.

Brent fell below $100 and reached roughly $97.76 during Tuesday trading as traders priced in the possibility of additional Middle Eastern supply. For financial markets, cheaper oil can become an important macroeconomic variable.

If sustained, lower crude prices could reduce some inflationary pressure, particularly for transportation and energy-intensive industries. That could influence expectations for interest rates and liquidity—the same variables that heavily affect risk assets such as Bitcoin.

Yet neither development eliminates uncertainty. Hormuz remains a geopolitical chokepoint, and Saudi pipeline capacity has not fully recovered. Bitcoin, meanwhile, still faces the volatility created by leveraged trading.

The larger story is therefore not simply Bitcoin at $87,000 or oil below $100. It is the interaction between capital flows, leverage, energy security and geopolitics. As institutional money returns to Bitcoin while energy-market fears temporarily ease.

Investors are watching whether this convergence can transform a powerful rebound into a more durable shift in global risk appetite.

Trump White House Media Ban Raises Questions About Press Freedom

By

Paul Ugbede Godwin

-

September 22, 2026

0

The relationship between the White House and the American press has entered a new and unusually consequential confrontation.

In September 2026, President Donald Trump’s administration barred journalists from CNN, MS NOW and Politico from entering the White House after Trump accused the organizations of publishing what he described as “fake news.”

The affected outlets have now taken the dispute to federal court, arguing that the restrictions violate constitutional protections for freedom of the press.

At the center of the dispute is a deceptively simple question: how much control should a president have over which journalists can enter the building where the executive branch conducts its business?

The White House argues that access is a privilege rather than an automatic right. In a statement published September 21, the administration said the First Amendment protects the ability of media organizations to publish, but does not guarantee them a press badge, briefing-room seat or position in the White House press pool.

The administration has also pointed to previous Democratic administrations and their disputes with conservative media as evidence that conflicts over presidential press access are not new.

The opposing argument is that the government cannot selectively restrict journalists because of unfavorable coverage. CNN, MS NOW and Politico allege that their exclusion represents viewpoint discrimination and violates their First and Fifth Amendment rights.

Their lawsuit seeks to restore their access to the White House. The significance extends beyond the three organizations involved. The White House press pool exists because no single news organization can place reporters everywhere a president goes.

Pool reporters share access, footage and information with other newsrooms, allowing television networks, newspapers and digital publications to cover presidential movements even when physical space is limited.

When CNN was removed from the television pool, the remaining members announced that they would stop participating rather than operate under the changed arrangement.

That creates a practical problem. Presidential communication is not simply another category of news. The White House makes decisions involving national security, foreign policy, markets, immigration and government spending.

Independent journalists provide questions, verification and competing accounts of those decisions. The administration, meanwhile, has increasingly emphasized direct communication with the public.

The White House launched an official app earlier this year promising real-time announcements, livestreams, photographs and administration updates directly to Americans. That approach allows the government to communicate without relying entirely on traditional media organizations.

This technological shift changes the economics of political information. Presidents can now speak directly to millions of people through official platforms and social media.

But direct communication and independent journalism serve different functions. An official White House statement explains what the administration wants the public to know; independent reporting can investigate, challenge and contextualize those claims.

The legal battle will therefore matter beyond the individual reporters who have lost access. It could help define how presidential administrations may manage press credentials, pool participation and physical access when relations between government officials and news organizations deteriorate.

For the moment, the conflict remains unresolved. A federal judge is considering the lawsuit filed by CNN, MS NOW and Politico.

What happens next could establish an important precedent for the relationship between the presidency and the press.

The White House may have greater authority to control its physical space, but the broader democratic question is whether that authority can be exercised according to the content or perceived fairness of a journalist’s reporting. That distinction is now being tested in court.

The Power and Danger of Concentration

By

Paul Ugbede Godwin

-

September 22, 2026

0

In June 2024, former OpenAI researcher Leopold Aschenbrenner published a 165-page essay arguing that artificial intelligence would require an extraordinary expansion of physical infrastructure.

The future of AI, he argued, would not be built only with better algorithms. It would require enormous quantities of chips, memory, electricity and data centers. He later turned that conviction into Situational Awareness, a hedge fund built around the same thesis.

The strategy was brutally concentrated. The fund bought companies positioned to benefit from the AI infrastructure boom while betting against software businesses that could be disrupted by increasingly capable artificial intelligence.

For a period, the results appeared extraordinary. According to the Financial Times, the fund reportedly returned 439 percent in the first half of 2026 and 1,551 percent since launch. By early July, it was managing approximately $45 billion.

Then the same concentration that created the extraordinary gains became a source of vulnerability. AI infrastructure stocks declined, while the software positions that had been expected to fall moved in the opposite direction.

Leverage amplified the problem. Prime brokers reportedly issued margin calls, forcing Situational Awareness to sell public-market positions to Ken Griffin’s Citadel at prices below their previous market levels. Within days, its assets reportedly fell to roughly $10 billion.

A subsequent letter to clients acknowledged that the fund had failed them. Yet the story did not end with retreat. Barely a week later, the fund was reportedly investing again, placing a $400 million bet.

That decision reveals something important: the underlying thesis and the financial structure supporting it are two different questions. Aschenbrenner may still be correct about the enormous infrastructure requirements of AI.

A correct long-term thesis can nevertheless produce devastating short-term losses when position sizes, leverage and liquidity become misaligned. That distinction extends far beyond hedge funds. Concentration is often the engine of exceptional performance.

Nobody becomes world-class by giving every skill equal attention. Psychologist Anders Ericsson and his colleagues’ research on deliberate practice emphasized sustained, purposeful effort directed toward specific areas of improvement.

Expertise requires repetition, feedback and depth. The person who spends years solving the hardest problems in a narrow field can develop capabilities that a generalist simply cannot reproduce. Career leverage works in much the same way.

If you want to become the person organizations call when a difficult problem appears, you need something unusually valuable to offer. That usually comes from going deeper than everyone else.

An engineer who understands a complex infrastructure system at an exceptional level, a journalist who develops unmatched expertise in a particular market, or a researcher who can connect technical developments with economic consequences creates scarcity around their knowledge.

But concentration has a boundary. The same specialization that accelerates expertise can become dangerous when it turns into total dependence. A career built around one employer, one technology, one customer, one market or one source of income can suffer the same problem as a highly leveraged investment portfolio.

When the central assumption breaks, there may be no protection. The lesson, therefore, is not to choose between concentration and diversification. It is to understand where each belongs. Concentrate your effort when building expertise.

Diversify your exposure when protecting what that expertise has created. Situational Awareness demonstrated both principles in spectacular fashion. Its concentrated thesis generated extraordinary returns, but its leverage and exposure also magnified the consequences when markets moved against it.

The goal is not simply to make a fortune from being right. It is to remain solvent, adaptable and useful long enough to benefit from being right again.

Nasdaq Hits Record as AI Stocks Rally, Oil Falls and Treasury Yields Ease

By

Samuel Nwite

-

September 22, 2026

0

US stocks rebounded sharply on Monday, with the Nasdaq Composite closing at a record high as investors returned to AI and semiconductor shares, while falling oil prices and lower Treasury yields eased two of the biggest pressures on equities.

The Nasdaq rose 2.26% to 27,122.09, its first record-high close since June 2. The S&P 500 gained 1.49% to 7,764.70, ending about 0.4% below its August 13 record, while the Dow Jones Industrial Average advanced 0.71% to 52,048.83.

The rally was led by semiconductor stocks. Intel surged 12.2%, Arm Holdings jumped 17%, and the PHLX Semiconductor Index gained 4.3%. Advanced Micro Devices climbed about 10%, pushing the chipmaker’s market value above $1 trillion for the first time.

The move suggests investors are again focusing on the earnings and spending cycle surrounding artificial intelligence after a sharp technology selloff the previous week triggered by warnings from executives and researchers about the risks of increasingly powerful AI systems.

AI-related companies have become a major source of earnings growth for the broader market, making semiconductor demand a particularly important indicator for investors trying to determine whether the AI investment cycle still has room to run.

Meta also joined the rally, with its shares jumping 11.4% after Wells Fargo raised its price target following the launch of the company’s Muse AI assistant.

The broader market received an additional boost from falling energy prices and Treasury yields. The benchmark 10-year Treasury yield moved back below 5%, while Brent crude futures fell below $100 a barrel for the first time in more than a week before settling at $100.34.

The decline in oil prices was linked to speculation that diplomatic efforts surrounding the Middle East conflict could make progress during the United Nations General Assembly in New York this week. US President Donald Trump said he would be open to meeting Iranian President Masoud Pezeshkian, who is expected to attend the gathering.

The shift in oil and bond markets mattered because both had recently been acting as significant constraints on equities. Higher crude prices raise concerns about inflation, while higher Treasury yields increase the discount rate applied to future corporate earnings and can make bonds more attractive relative to stocks.

“We’ve been kind of programmed to follow the price of oil and the yield on the US 10-year, and if you look at those two things today, they’ve shifted from being headwinds to tailwinds for this market, at least in the near term,” said Art Hogan, chief market strategist at B. Riley Wealth.

The improvement in those two markets, however, does not eliminate the Federal Reserve’s inflation problem. The central bank raised its benchmark interest rate last week for the first time in three years, and traders now see roughly a 50% probability of another increase next month, according to CME’s FedWatch tool.

That leaves the market sensitive to economic data and comments from policymakers. At least 10 Federal Reserve officials are scheduled to speak this week, giving investors additional clues about whether the central bank intends to continue tightening monetary policy.

Valuations also remain an important consideration. The S&P 500 was trading at just under 19 times expected earnings on Friday, according to LSEG data, its lowest forward valuation since 2023. At the same time, much of the improvement in earnings expectations has been concentrated among AI-related technology companies.

That creates a tension in the current rally. Lower valuations can provide support for equities, but the market’s earnings outlook is increasingly dependent on whether the AI investment boom continues to generate revenue and profits at a pace that justifies the enormous spending on chips, data centers and computing infrastructure.

Monday’s semiconductor rally suggests investors remain willing to finance that thesis. AMD’s move above a $1 trillion market capitalization was particularly significant because it signals how strongly investors are valuing companies positioned to benefit from continued AI infrastructure spending.

Other themes also contributed to the market’s risk-on tone. Bitcoin rose more than 6% to a seven-month high, lifting shares of Coinbase and Strategy. US-listed companies with exposure to Greenland also surged after progress in diplomatic discussions between Washington and Copenhagen. Greenland Mines tripled, while Greenland Energy more than doubled.

Investors are also watching the expected US-China summit later this week. Discussions could include an extension of the two countries’ trade truce, the Middle East conflict and AI regulation. A report said Washington had proposed extending the truce for six months, while Beijing was seeking a longer extension.

The gains were not universal. Paramount Skydance fell nearly 3% after Reuters reported that the company had reached a settlement with California and 11 other states that had sued to block its proposed $110 billion acquisition of Warner Bros. Discovery.

Eight of the 11 S&P 500 sector indexes finished higher. Communication services led the advance with a 4.16% gain, followed by information technology, which rose 2.4%.

Market breadth was more mixed than the headline indexes suggested. Advancing stocks outnumbered declining stocks in the S&P 500 by only 1.4 to one. The S&P 500 recorded seven new highs and 29 new lows, while the Nasdaq posted 64 new highs against 127 new lows.

Trading volume was relatively strong, with 16.5 billion shares changing hands on US exchanges, above the 20-session average of 16.2 billion.

Monday’s rally thus amounted to more than a simple rebound in technology stocks. Investors simultaneously received relief from lower oil prices and Treasury yields, while semiconductor gains reinforced the view that AI spending remains a powerful source of corporate earnings growth.

The key question for markets is how long those conditions will last. With the Fed still focused on inflation and AI valuations increasingly dependent on continued infrastructure spending, the latest record for the Nasdaq has strengthened the market’s momentum without removing the forces that could challenge it.