Key Highlights

● VNET Group's wholesale data-center capacity crossed 1 gigawatt at 1,007 MW with utilization near 74%, even as second-quarter 2026 net Revenue of RMB2.78 billion came with a net loss of RMB135.6 million.

● Kingsoft Cloud posted its first GAAP operating profit of RMB23.0 million in the second quarter of 2026 as revenue rose 82% year over year to RMB3,072.0 million.

● Cognizant's trailing-12-month bookings reached $29.1 billion, up 5%, while second-quarter 2026 revenue grew 4.5% to $5,481 million and adjusted diluted EPS rose 4.6% to $1.37.

● Meta Platforms grew second-quarter 2026 revenue 28% to $60.8 billion but saw Operating Margin compress to 31% as Capital Expenditure more than doubled to $31.1 billion and full-year 2026 capex guidance reached $130 to $145 billion.

Five Nasdaq-listed names, a Chinese data-centre operator, a cloud challenger, a biometric micro-cap, a global IT-services firm and a social-media giant, trace the full arc of the AI trade as the Federal Reserve prepares its September decision.

A Single Index, Five Very Different Bets on Artificial Intelligence

The Nasdaq Composite has never been a monolith, and September 2026 makes that plain. On paper, VNET Group (NASDAQ: VNET), Kingsoft Cloud (NASDAQ: KC), BIO-key International (NASDAQ: BKYI), Cognizant (NASDAQ: CTSH) and Meta Platforms (NASDAQ: META) share only an exchange. In practice they map the entire AI value chain: the power and racks that host the models, the cloud layer that rents them out, the identity software that guards access, the services firms that stitch AI into enterprise workflows, and the platform spending tens of billions to build frontier systems.

They also span the risk spectrum, from a sub-$3 micro-cap to a company worth more than a trillion dollars. Reading them together, on the eve of a Federal Reserve meeting that could go either way, is a useful way to gauge how much of the AI story is already priced in.

A Hawkish Backdrop That Complicates the AI Trade

The timing matters. The Federal Open Market Committee meets on 15 and 16 September 2026, with a decision due the 16th, two days after this analysis was compiled.

According to market commentary, the federal funds target sits at 3.50% to 3.75%, and the debate is unusually lopsided toward tightening. Fed Chair Kevin Warsh's Jackson Hole remarks in late August pushed the perceived probability of a 25-basis-point hike above 50%, up from roughly 36% beforehand, as officials signalled they want clearer evidence of sustained disinflation before easing. A firm early-September jobs report reignited those hike fears, and market coverage noted the Dow falling more than 260 points on 3 September as the data landed.

That is an awkward setting for long-duration technology equities, whose valuations lean on distant cash flows. It sharpens the divide between AI names that are already converting spending into cash and those still asking investors for patience. The five companies below sit on both sides of that line.

VNET Group (NASDAQ:VNET): Powering China's AI Build-Out, One Gigawatt at a Time

VNET Group, the carrier-neutral data-centre operator formerly known as 21Vianet, is a direct play on China's Demand for AI compute capacity.

The company's second-quarter 2026 results, reported in USD using its own conversion, put net revenue at RMB2.78 billion ($135.3 million), up 25.4%, with a 33.0% margin, though the company still posted a net loss of RMB135.6 million.

The headline operational milestone: wholesale capacity in service crossed 1 gigawatt, at 1,007 MW, with 970 MW committed and utilisation near 74%. The company also flagged a "compute-energy" Partnership with battery maker CATL.

The shape here is classic infrastructure: heavy capex now, guided at RMB10 to 12 billion for 2026, losses today, and EBITDA and committed Backlog as the forward tells. The risk is that this is a capital-intensive, China-exposed bet dependent on demand the company cannot fully control.

Kingsoft Cloud (NASDAQ:KC): The Quarter the AI Cloud Finally Paid

Kingsoft Cloud, also listed on the Hong Kong exchange, offers the cleaner momentum story of the two Chinese names.

Second-quarter 2026 total revenue was RMB3,072.0 million ($195.6 million), up 82% year over year and now 56% of public cloud revenue. Most notably, the company reported GAAP operating profit of RMB23.0 million, described as its first positive operating quarter, and narrowed its net loss by 79.6% to RMB93.0 million. Non-GAAP EBITDA margin reached 35.8%.

Kingsoft Cloud is where VNET hopes to be: the AI infrastructure spend is translating into revenue that is beginning to carry its own weight. The reliance on the broader Kingsoft and Xiaomi ecosystem cuts both ways, a captive demand base, but also a concentration risk. One quarter of operating profit is a marker, not a trend; sustaining it through continued capex is the open question.

BIO-key International (NASDAQ:BKYI): A Biometric Micro-Cap That Survived a Delisting Scare

BIO-key International is the smallest and most speculative name here, a biometric and identity-access-management software vendor whose story in 2026 has been as much about its listing as its products.

The verification detail matters. The company executed a 1-for-10 reverse split effective 30 April 2026, was suspended from Nasdaq and moved to OTC markets on 12 to 13 May over a minimum-bid-price shortfall and a late annual filing, then regained compliance: following a June Hearings Panel decision, trading resumed on the Nasdaq Capital Market on 8 July 2026. It remains under a discretionary panel monitor through mid-2027. So the premise holds, the company is again Nasdaq-listed, but only just.

On the business, second-quarter 2026 revenue rose 13% to $1.92 million, gross margin expanded to 87% from 73%, and the net loss narrowed to roughly $577,000. License revenue grew 53%, with wins cited across government, defence and financial customers in Portugal, Jordan and the US. Management targets profitability in the second half. The identity-security theme is real, but at this scale execution and going-concern math dominate; any forward optimism should be treated cautiously.

Cognizant (NASDAQ:CTSH): The IT-Services Test Case for Whether AI Is a Headwind or a Tailwind

Cognizant is the bellwether for how AI reshapes the outsourcing and systems-integration business.

Second-quarter 2026 revenue was $5,481 million, up 4.5% (4.1% in constant currency), with GAAP operating margin of 15.9% and adjusted diluted EPS of $1.37, up 4.6%. Trailing-12-month bookings reached $29.1 billion, up 5%, and the company repurchased $1,153 million of stock. Full-year 2026 guidance calls for revenue of $22.04 to $22.35 billion and adjusted EPS of $5.70 to $5.82. Management framed the company as an "AI builder" scaling a "Frontier workforce" to close clients' "AI velocity gap."

This is the crux of the services debate: does generative AI cannibalise billable hours, or does it create a wave of modernisation, data and integration work that firms like Cognizant capture? Mid-single-digit growth with expanding margins and a buyback suggests a company managing the transition rather than being run over by it, but not yet accelerating. The AI-services thesis needs bookings to convert into faster revenue before the market re-rates it.

Meta Platforms (NASDAQ:META): The Mega-Cap Where AI Ambition Meets Margin Reality

Meta Platforms is the gravitational centre of this group and the sharpest illustration of AI's cost.

Second-quarter 2026 revenue was $60.8 billion, up 28% year over year, with Advertising at $59.4 billion. Yet Net Income was $15.8 billion and diluted EPS of $6.18 came in below analyst expectations near $7.17, as operating margin compressed to 31%. The reason is spending: quarterly capital expenditure hit $31.1 billion, more than double the prior year, total expenses rose 55% to $42.0 billion, and Reality Labs lost $4.6 billion. The company guided full-year 2026 capex to $130 to $145 billion. Reporting noted shares fell roughly 9.6% after hours.

The advertising machine is still humming, but the market is being asked to underwrite an enormous, front-loaded Investment in AI infrastructure and, ultimately, "superintelligence." The bull case is that today's spend builds a durable moat and future revenue; the bear case is that returns stay distant while margins keep shrinking. This is the highest-conviction AI bet of the five, and the one where the payoff timeline is most contested.

What Investors Should Watch

For the two China names, the tell is the same: whether AI-driven demand keeps converting into cash. Watch VNET's committed-versus-utilised capacity and its heavy capex burn, and watch whether Kingsoft Cloud can repeat operating profitability rather than log a single quarter.

For Cognizant, the number that matters is bookings momentum and any sign that AI is accelerating, not eroding, services demand. For BIO-key, the binary questions are the second-half profitability target and continued Nasdaq compliance under its panel monitor.

For Meta, and for the index overall, the September FOMC decision is the near-term swing factor. A hawkish surprise would pressure exactly the long-duration, capex-heavy AI names that have led the market. Watch the reaction on 16 September.

Spending, Revenue and Profit Are Three Different Stories Here

Kingsoft Cloud's first operating profit and Cognizant's steady margins sit alongside VNET's losses, BIO-key's survival math and Meta's margin compression, and that spread is the real story of this basket. Together they show an AI trade that has clearly moved from promise to spending, and in places from spending to revenue, but not uniformly to profit.

With the Fed leaning hawkish into its September meeting, the market's tolerance for invest-now, earn-later strategies is being tested in real time. What separates these five is not enthusiasm but arithmetic: whether committed capacity, bookings and license wins are actually turning into cash before the capital comes due again next quarter.