About $12 billion flowed out of U.S.-listed gold and Bitcoin exchange-traded funds in the first half of 2026, while semiconductor-focused ETFs attracted more than $20 billion in net inflows, according to multiple market data sources. The shift suggests that traders have not abandoned risk assets altogether. Instead, capital has rotated toward artificial intelligence infrastructure, chipmakers and data-center supply chains.
The change has been most visible in spot Bitcoin ETFs, which suffered their weakest month since launch in June. The products recorded roughly $4.5 billion in net outflows during the month, pushing cumulative flows into negative territory. Gold ETFs also lost capital as traders moved away from traditional stores of value and toward companies tied to the physical expansion of AI computing capacity.
At the same time, five major U.S. technology companies are expected to spend between $600 billion and $725 billion on AI infrastructure this year, with about 70% of that total directed toward chips, servers and data centers. That spending wave has made semiconductor ETFs one of the clearest beneficiaries of the market’s current preference for tangible computing assets over financial instruments such as digital tokens or bullion-backed funds.
The rotation has created a sharper divide inside the broader risk market. Bitcoin remains highly sensitive to liquidity, interest-rate expectations and ETF demand. AI-linked hardware, by contrast, is being supported by multiyear contracts, corporate capital budgets and urgent demand for computing power. That difference has made the semiconductor trade look more durable to some traders, even as concerns grow that valuations in the sector may be crowded after months of strong gains.
Semiconductors pull liquidity from gold and Bitcoin
The first-half ETF data show a market that is reallocating rather than simply de-risking. Capital has left gold funds and Bitcoin products, but it has not moved entirely into cash or low-risk bonds. Instead, traders have favored semiconductor ETFs, which are linked to the chips and memory systems required to train and run AI models.
That distinction is important because it shows how risk appetite has changed. Earlier cycles often saw Bitcoin benefit from loose liquidity, low borrowing costs and enthusiasm for high-growth technology themes. In 2026, the AI infrastructure boom has absorbed a large share of that same speculative and growth-oriented capital.
The market preference is also more physical than in previous digital asset cycles. The leading AI trade is tied to factories, data centers, power systems, networking equipment and advanced chips. Cash is being committed to projects with long construction timelines and payback periods that may stretch over several years. That leaves less money available for assets that move mainly on liquidity expectations and short-term flows.
Component suppliers are already under pressure to meet demand. Market data cited by industry participants show that AI-related technology spending has consumed a dominant share of global computer memory output for the fiscal year. The competition for high-bandwidth memory, advanced graphics processors and server capacity has tightened supply chains and helped sustain strong demand for semiconductor exposure.
This physical buildout has also changed how traders view relative opportunity. Bitcoin offers liquidity, volatility and scarcity narratives. AI infrastructure offers revenue visibility through long-term service agreements and corporate spending plans. In the current market, the second story has attracted more capital.
Bitcoin ETFs suffer their weakest month
June was a difficult month for spot Bitcoin ETFs. Net outflows of about $4.5 billion marked the weakest monthly performance since the products were introduced, and the withdrawals were large enough to turn cumulative flows negative.
The pressure did not end with June. Fresh data from July 9 showed that spot Bitcoin funds lost another $95.3 million in a single trading session. Products managed by Fidelity and Ark Invest were among those with the largest withdrawals, adding to evidence that institutional capital remained cautious toward Bitcoin exposure even after prices stabilized.
Bitcoin did rebound above $63,000 in early July after spot ETF flows briefly turned positive. But the recovery remains fragile because ETF demand has become a key gauge of marginal buying pressure. When these products attract steady inflows, they can support spot prices by absorbing available supply. When they record persistent outflows, they can deepen selling pressure or limit rallies.
Traders are also watching the $57,000 region as a possible support level if renewed weakness develops. Some market participants view that area as a more attractive zone for reentry if the current capital rotation continues. However, sustained improvement in flows is likely to matter more than any single price level.
The central question is whether June’s weakness was a temporary reset or part of a longer shift in capital allocation. Early July’s brief positive flows offered some relief, but one or two sessions are unlikely to confirm a durable trend. Many traders are looking for several consecutive days of net inflows before concluding that demand has returned in a meaningful way.
On-chain supply signals remain supportive
Despite weak ETF flows, some Bitcoin supply indicators remain constructive. Exchange reserves have dropped to their lowest level in roughly seven years, according to market data cited by digital asset researchers. Lower exchange balances can indicate that fewer coins are immediately available for sale, although the signal is not always enough to offset weak demand from ETFs or broader macro pressure.
Long-term holders also continue to accumulate at one of the fastest rates seen across recent cycles. This suggests that core Bitcoin holders have not broadly abandoned the asset, even as short-term ETF flows have deteriorated.
That creates a mixed picture. On one side, supply held on trading venues is shrinking, and long-term accumulation remains strong. On the other side, spot ETF outflows show that recent demand from regulated fund products has weakened. For Bitcoin to regain stronger momentum, traders will likely need to see both supply tightness and renewed inflows working in the same direction.
The July rebound above $63,000 showed that the market can recover quickly when ETF flows turn less negative. But the move also highlighted how dependent Bitcoin has become on financial-market liquidity. Without a broader return of capital to crypto-linked products, price gains may remain vulnerable to reversals.
AI buildout reshapes crypto mining
The capital shift is also changing the business model of Bitcoin mining companies. Several miners are repurposing facilities to provide computing power for AI clients, using their access to energy, land and high-performance infrastructure to enter the data-center market.
TeraWulf and other mining-linked companies have benefited from this trend. TeraWulf posted returns of about 73% in 2026, reflecting stronger market interest in miners that can reposition themselves as AI infrastructure providers. Estimates cited by market analysts suggest that nearly 70% of revenue for listed miners could come from AI contracts by the end of the year if current trends continue.
The change is significant because mining companies have historically depended on Bitcoin prices, block rewards and energy costs. After the latest Bitcoin halving, miners faced tighter economics and stronger pressure to diversify. AI hosting gives them another source of revenue, often tied to longer contracts and more predictable demand.
This does not mean every miner can successfully make the transition. AI workloads require different technical standards, stronger uptime commitments, advanced cooling systems and reliable power delivery. Companies with strong balance sheets and suitable infrastructure may benefit, while weaker operators may struggle to fund the upgrades needed.
Still, the trend shows how AI is drawing capital and strategic attention even from companies originally built around Bitcoin. Rather than waiting only for the next crypto cycle, some miners are turning directly toward the sector currently receiving the most funding.
Structural shift or cyclical rotation
The biggest debate is whether the move away from Bitcoin and gold ETFs is structural or cyclical. Executive Chairman Michael Saylor has described the pattern as a cyclical rotation, implying that capital may eventually return to Bitcoin once the AI trade cools or becomes too expensive.
Others are more cautious. If AI infrastructure has entered a multiyear capital cycle, the return of liquidity to Bitcoin could take much longer than in past rotations. Spending on chips, data centers and power systems is not as easy to reverse as a short-term trade. Once large companies commit to construction projects and supply agreements, capital can remain locked into those plans for years.
That makes the timing of any rotation back toward digital assets uncertain. Money allocated to infrastructure with long payback horizons may not quickly return to liquid markets, even if Bitcoin prices become more attractive. The capital cycle in AI may need to show signs of slowing before traders broadly reconsider crypto exposure.
At the same time, the AI trade carries its own risks. Valuations have risen sharply, and some technology shares are priced for sustained growth. If upcoming earnings reports show margin pressure, slowing orders or rising infrastructure costs, enthusiasm for semiconductor exposure could fade. That would create conditions for some capital to rotate back into Bitcoin and other high-volatility assets.
Earnings season becomes a key test
Late July technology earnings reports are expected to be closely watched for evidence that the AI infrastructure boom remains profitable. Traders are likely to focus less on headline growth and more on whether companies can turn heavy spending into sustainable earnings.
Margins will be especially important. Hardware makers and data-center operators are spending aggressively to secure chips, land, electricity and cooling capacity. If revenue growth does not keep pace with those costs, the market may begin to question whether AI-related valuations have moved too far.
Analyst Rommel said the broader market focus is shifting from raw growth toward the durability of future earnings. That view reflects a wider concern that AI leaders must now prove they can deliver strong returns on massive capital spending programs.
If chipmakers and cloud companies meet or exceed expectations, semiconductor ETFs could continue drawing inflows. If results disappoint, the infrastructure trade may lose momentum, especially after months of heavy positioning. In that scenario, Bitcoin could benefit from a partial rotation if traders seek other high-volatility assets with depressed flows.
Semiconductor sales figures scheduled for August 15 will also provide another important signal. Those numbers may show whether intense demand for chips and memory is beginning to level off or whether supply shortages are likely to last through the end of the calendar year.
Higher rates tighten the risk environment
Macro conditions remain a major challenge for both Bitcoin and high-growth technology assets. Deutsche Bank projects two additional U.S. Federal Reserve interest-rate hikes in 2026. Higher borrowing costs can weigh on speculative assets by raising the return available from cash and short-term debt while making future growth less valuable in present terms.
For Bitcoin, higher rates are often a headwind because the asset does not generate yield. When borrowing costs rise, traders may demand stronger price momentum or clearer catalysts before allocating capital to digital assets.
For AI infrastructure, higher rates can raise financing costs for data centers, power projects and chip production. However, the sector has so far benefited from the size and urgency of corporate spending plans. Large technology firms with strong cash flows may be better positioned to absorb higher rates than smaller speculative companies.
A weak jobs report or cooler inflation data in mid-August could quickly alter expectations. If markets begin pricing in lower rates or a pause in tightening, non-yielding assets such as Bitcoin and gold could become more attractive again. Lower borrowing costs would also ease financing pressure across high-growth sectors.
Regulatory and local risks grow
The AI infrastructure boom is not only facing market and earnings risks. Local political resistance to new data centers has become a growing obstacle. Community concerns over electricity demand, water usage and land development have already delayed or blocked major projects across the United States.
Market participants estimate that local pushback has affected as much as $130 billion worth of planned technology facilities. If resistance grows, the buildout of AI infrastructure could slow, forcing companies to revise timelines or move projects to regions with more supportive energy and permitting conditions.
That risk is important because the semiconductor trade depends not only on chip demand but also on the ability to deploy those chips at scale. Advanced processors require enormous data-center capacity, reliable power and cooling. If infrastructure bottlenecks intensify, some of the expected spending could be delayed.
Bitcoin faces a different set of regulatory questions. ETF demand remains tied to broader acceptance of digital assets within traditional financial markets, while miners continue to face scrutiny over energy use. The overlap between AI data centers and mining facilities may draw even more attention from local officials and regulators as both industries compete for power.
Liquidity now moves within one risk pool
The first-half flow data show that Bitcoin, gold and semiconductor ETFs are increasingly connected through the same global liquidity cycle. Capital is not simply moving from risky assets to safe assets. It is rotating within a shared pool of volatile and growth-sensitive trades.
Gold and Bitcoin have lost ground as traders favor AI-linked hardware. Semiconductor ETFs have gained as corporate spending on chips and data centers accelerates. Mining companies are adapting by shifting toward AI hosting. Meanwhile, Bitcoin’s long-term supply metrics remain firm, but ETF outflows show that immediate demand is still uneven.
The next phase will likely depend on three signals: whether Bitcoin ETF flows can remain positive for more than a few sessions, whether AI earnings support current valuations, and whether macro data reduces pressure from higher interest rates.
For now, the market is showing a clear preference for concrete, silicon and computing capacity over digital scarcity. That does not mean Bitcoin’s role in the risk market has disappeared. It means the asset is competing with one of the largest infrastructure spending cycles in modern technology, and that competition is shaping where liquidity moves next.
Track shifting liquidity between Bitcoin and AI sectors using real-time ETF and crypto data on Toobit Markets.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
