Voss Capital's 22.4% quarter beat the Russell 2000 by 0.9 points on a book 100.6% net long
Voss Capital's Value Fund returned 22.4% net in Q2 2026, 0.9 points ahead of the Russell 2000's 21.5%. With the master fund 100.6% delta-adjusted net long on June 30, we think the small-cap tape explains most of that gain.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+15
- Incentives60
- Confidence65
Jefferies' Chris Wood calls much of the hyperscalers' AI spending malinvestment
Jefferies' Chris Wood says US hyperscalers' $695 billion of guided 2026 capex looks unsustainable as cheap Chinese open-source models win share. He blames a slump in the price of AI output, with demand intact, while guidance climbs toward $870 billion for 2027.
Reality
- Evidence30
- Adoption45
- Hype gap+35
- Incentives
- Not scored
- Confidence35
US chipmakers' expected profit growth slows to 136%, more than four times the S&P 500's pace
LSEG expects US chipmakers' third-quarter profit growth to slow to about 136% from 158% while S&P 500 earnings rise about 31%. Analysts have raised that quarter's forecast at half their earlier pace since July, so the AI earnings carrying the index are losing speed.
Reality
- Evidence55
- Adoption65
- Hype gap+10
- Incentives30
- Confidence55
buildConfirmed2 publishers Andreessen Horowitz puts hyperscaler capex near $780 billion this year and tech at roughly 55% of US capital spending. Its charts also show free cash flow being squeezed, and the hyperscalers are relying more on debt to fund new chip capacity.
Reality
- Evidence55
- Adoption30
- Hype gap+20
- Incentives75
- Confidence60
SemiAnalysis's count of more than 1,000 Chinese datacenters puts the operating fleet above 24GW, larger than EMEA or the rest of Asia. For listed landlords such as GDS and VNET, the open question moves from how big the market is to how much rent its largest tenants will pay.
Reality
- Evidence45
- Adoption60
- Hype gap+20
- Incentives70
- Confidence50
Goldman sees the five biggest hyperscalers adding $400 billion of AI capex in 2027
Goldman Sachs expects the five biggest US hyperscalers to spend about $1.2 trillion on AI infrastructure in 2027, 50% more than in 2026. The forecast is $100 billion above consensus, and the bank lists power, labor and memory chips as the limits on building it.
Reality
- Evidence35
- Adoption55
- Hype gap+25
- Incentives
- Not scored
- Confidence40
Every net dollar goes to AI infrastructure, three days after quarterly profit fell 75%. For anyone buying cloud AI, that changes which number tells you whether capacity arrives.
Reality
- Evidence72
- Adoption55
- Hype gap+20
- Incentives70
- Confidence65
Computer equipment outspent US housing by $4 billion in the second quarter
Real spending on information processing equipment reached $752 billion against $748 billion of residential investment. The gap is half a percent, and the six-company capex forecast that supports the concentration argument is a separate series.
Reality
- Evidence62
- Adoption52
- Hype gap+33
- Incentives38
- Confidence58
Converting miners' 3,450MW of contracted but not-yet-billing AI capacity could cost $28bn to $52bn
CoinShares marks stabilised data centre capacity at $27m a megawatt and energised mining capacity at $3m, and the conversion cost in between explains why Core Scientific, Keel, Cipher and IREN are walking away from hashrate.
Reality
- Evidence45
- Adoption38
- Hype gap+35
- Incentives70
- Confidence45
buildOne report1 publisher Amazon moved a subset of its servers back to a five-year depreciation life
Amazon's filing blamed the pace of change in artificial intelligence and machine learning, and it priced the decision itself: about $0.7 billion off 2025 operating income, with another $0.6 billion of accelerated depreciation.
Reality
- Evidence62
- Adoption68
- Hype gap+12
- Incentives72
- Confidence58
Oracle answers its surprise AI bill by showing employees what each model costs
Oracle put ChatGPT Enterprise and OpenAI's Codex in front of about 160,000 employees in April, and 80 percent were using them within three months. Then the bills arrived and, the CIO said, surprised the company.
Reality
- Evidence40
- Adoption68
- Hype gap+15
- Incentives68
- Confidence46
Morgan Stanley puts the last groundbreaking for 2027 data-center capacity in October
Build times of two to three years make an October start the last one that can open before 2028, and JPMorgan already counts more than 60 percent of planned 2027 capacity as unbuilt. The scarce input is a permitted site with power.
Reality
- Evidence28
- Adoption
- Not scored
- Hype gap+30
- Incentives55
- Confidence34
The €13 billion Google is putting into Finnish data centres in 2027 and 2028 comes with a two-decade electricity contract that Fortum says will support its programme to extend the Loviisa reactors' operating life.
Reality
- Evidence60
- Adoption52
- Hype gap+15
- Incentives74
- Confidence66
Oracle spent $28bn of cash in the three months it reported $4.68bn of net income
Oracle's first fiscal quarter turned 30% revenue growth into a 60% jump in net income, and the same three months consumed more cash in capital investment than the company collected in sales.
Reality
- Evidence52
- Adoption70
- Hype gap+18
- Incentives72
- Confidence58
Microsoft's $19.4bn Nebius contract works out to under $194,000 per GB300
The deal is the largest slice of more than $33bn Microsoft has committed to specialist GPU clouds. It leaves the fleet, the debt behind it and the 18-month depreciation clock on someone else's balance sheet.
Reality
- Evidence27
- Adoption46
- Hype gap+38
- Incentives66
- Confidence31
The Fed's Beige Book now rests US construction on data center capex
PwC puts global AI infrastructure spending at $31.6 trillion through 2050, but its own annual path compounds at only about 3.4 percent a year, which makes the near-term dependence sharper than the headline total does.
Reality
- Evidence38
- Adoption55
- Hype gap+30
- Incentives68
- Confidence44
PwC's 2050 model has data centres spending 93 cents of every capex dollar on equipment
PwC's Oxford Economics modelling puts $31.6 trillion into AI infrastructure by 2050 and raises equipment to 93% of data centre capex, which is the figure that decides how any of these sites get financed.
Reality
- Evidence38
- Adoption
- Not scored
- Hype gap+18
- Incentives72
- Confidence44
With 2,059 more US data centers planned and three-quarters of Americans opposed to having one near them, local consent is now the part of the buildout that committed capital cannot put on a calendar.
Reality
- Evidence64
- Adoption71
- Hype gap+34
- Incentives76
- Confidence66
Moody's puts China's 2026 AI build at $140 billion against $785 billion for six US hyperscalers
The rating agency reads the gap as cash flow rather than ambition, but its own report ranks chip access as the tighter constraint, and the June quarter run rate already sits above the 2026 forecast.
Reality
- Evidence47
- Adoption71
- Hype gap+14
- Incentives44
- Confidence56
Electrical technician postings outran the wider job market by 45 to one
PwC has US data center spending more than doubling by 2027. Deloitte's posting data shows crews arriving from outside the trades, and mentoring capacity is quietly the input that decides the schedule.
Reality
- Evidence42
- Adoption
- Not scored
- Hype gap+24
- Incentives58
- Confidence40